github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-15 12:48:05 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #1289] llama2 and GPU execution on Nvidia Jetson #664

New Issue
Closed
opened 2026-04-12 10:21:12 -05:00 by GiteaMirror · 16 comments
GiteaMirror commented 2026-04-12 10:21:12 -05:00
Owner
Copy Link

Originally created by @dillera on GitHub (Nov 27, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1289

I'm trying to use ollama, llama2 and the nvida jetson.

@bnodnarb made a nice tutorial just a bit ago here:
https://github.com/jmorganca/ollama/pull/1098/files

I've been trying this without success, I can't get llama to use the gpu on the jetson. I'm hoping bnodnarb is around and can discuss a bit more about what was done to make this work w/ the GPU.

Originally created by @dillera on GitHub (Nov 27, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1289 I'm trying to use ollama, llama2 and the nvida jetson. @bnodnarb made a nice tutorial just a bit ago here: https://github.com/jmorganca/ollama/pull/1098/files I've been trying this without success, I can't get llama to use the gpu on the jetson. I'm hoping bnodnarb is around and can discuss a bit more about what was done to make this work w/ the GPU.
GiteaMirror commented 2026-04-12 10:21:12 -05:00
Author
Owner
Copy Link

@bnodnarb commented on GitHub (Nov 28, 2023):

Hi @dillera - How much memory does your Jetson have? The Orin Nano dev kit comes with 8GB, which is enough to run many of the 7B models. The regular Jetson Nano (non Orin) comes with either 2GB or 4GB, which is only enough to run some of the smaller models you might find on Hugging Face.

Do you think the issue you're facing might be related to this?

<!-- gh-comment-id:1828875584 --> @bnodnarb commented on GitHub (Nov 28, 2023): Hi @dillera - How much memory does your Jetson have? The Orin Nano dev kit comes with 8GB, which is enough to run many of the 7B models. The regular Jetson Nano (non Orin) comes with either 2GB or 4GB, which is only enough to run some of the smaller models you might find on Hugging Face. Do you think the issue you're facing might be related to this?
GiteaMirror commented 2026-04-12 10:21:13 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

Hey thanks for replying. I got the 8GB one.

I've followed your directions and I never see a blip on GPU jtop or the PowerGUI- it just runs on the CPUs. I even went in and modified the systemd service file to set that env variable you were doing with your tee so that ollama service just always started with that...

All I get from logs is:

Nov 27 16:54:54 jetson ollama[7008]: NvRmMemInitNvmap failed with Permission denied
Nov 27 16:54:54 jetson ollama[7008]: 549: Memory Manager Not supported
Nov 27 16:54:54 jetson ollama[7008]: ****NvRmMemInit failed**** error type: 196626
Nov 27 16:54:54 jetson ollama[7008]: *** NvRmMemInit failed NvRmMemConstructor
Nov 27 16:54:54 jetson ollama[7008]: CUDA error 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported
Nov 27 16:54:54 jetson ollama[7008]: current device: 0
Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:435: 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported
Nov 27 16:54:54 jetson ollama[7008]: current device: 0
Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:443: error starting llama runner: llama runner process has terminated
Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:509: llama runner stopped successfully
Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:420: starting llama runner
Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:478: waiting for llama runner to start responding
Nov 27 16:54:54 jetson ollama[11241]: {"timestamp":1701122094,"level":"WARNING","function":"server_params_parse","line":873,"message":"Not compiled with GPU offload support, --n-gpu-layers option will be ignored. See main README.md for information on enabling GPU BLAS support","n_gpu_layers":-1}
Nov 27 16:54:54 jetson ollama[11241]: {"timestamp":1701122094,"level":"INFO","function":"main","line":1324,"message":"build

Now, to be fair I'm using llama2 and not mistral, but it shouldn't matter on the model should it?

Startup:

dillera@jetson:~$ cat /etc/systemd/system/ollama.service 
[Unit]
Description=Ollama Service
After=network-online.target

[Service]
ExecStart=/usr/local/bin/ollama serve
User=ollama
Group=ollama
Restart=always
RestartSec=3
Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64"
Environment="PATH=/usr/local/cuda/lib64:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin"

[Install]
WantedBy=default.target

I'm now messing around with the docker container for llama.ccp here:
https://github.com/dusty-nv/jetson-containers/tree/master/packages/llm/llama_cpp

but using ollama is a bit easier than messing with llama's main directly.

I got this thing to see what it would do with this GPU, it's sort of lame that it's so hard to use it!

What I run once ollama is started:


ollama pull llama2

ollama create llama2-jetson -f ./ModelfileLlama2Jetson

ollama run llama2-jetson

and

dillera@jetson:~$ cat ./ModelfileLlama2Jetson 
FROM llama2
PARAMETER num_gpu 999

thanks!

<!-- gh-comment-id:1829006115 --> @dillera commented on GitHub (Nov 28, 2023): Hey thanks for replying. I got the 8GB one. I've followed your directions and I never see a blip on GPU jtop or the PowerGUI- it just runs on the CPUs. I even went in and modified the systemd service file to set that env variable you were doing with your tee so that ollama service just always started with that... All I get from logs is: ``` Nov 27 16:54:54 jetson ollama[7008]: NvRmMemInitNvmap failed with Permission denied Nov 27 16:54:54 jetson ollama[7008]: 549: Memory Manager Not supported Nov 27 16:54:54 jetson ollama[7008]: ****NvRmMemInit failed**** error type: 196626 Nov 27 16:54:54 jetson ollama[7008]: *** NvRmMemInit failed NvRmMemConstructor Nov 27 16:54:54 jetson ollama[7008]: CUDA error 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported Nov 27 16:54:54 jetson ollama[7008]: current device: 0 Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:435: 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported Nov 27 16:54:54 jetson ollama[7008]: current device: 0 Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:443: error starting llama runner: llama runner process has terminated Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:509: llama runner stopped successfully Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:420: starting llama runner Nov 27 16:54:54 jetson ollama[7008]: 2023/11/27 16:54:54 llama.go:478: waiting for llama runner to start responding Nov 27 16:54:54 jetson ollama[11241]: {"timestamp":1701122094,"level":"WARNING","function":"server_params_parse","line":873,"message":"Not compiled with GPU offload support, --n-gpu-layers option will be ignored. See main README.md for information on enabling GPU BLAS support","n_gpu_layers":-1} Nov 27 16:54:54 jetson ollama[11241]: {"timestamp":1701122094,"level":"INFO","function":"main","line":1324,"message":"build ``` Now, to be fair I'm using llama2 and not mistral, but it shouldn't matter on the model should it? Startup: ``` dillera@jetson:~$ cat /etc/systemd/system/ollama.service [Unit] Description=Ollama Service After=network-online.target [Service] ExecStart=/usr/local/bin/ollama serve User=ollama Group=ollama Restart=always RestartSec=3 Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64" Environment="PATH=/usr/local/cuda/lib64:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin" [Install] WantedBy=default.target ``` I'm now messing around with the docker container for llama.ccp here: https://github.com/dusty-nv/jetson-containers/tree/master/packages/llm/llama_cpp but using ollama is a bit easier than messing with llama's main directly. I got this thing to see what it would do with this GPU, it's sort of lame that it's so hard to use it! What I run once ollama is started: ``` ollama pull llama2 ollama create llama2-jetson -f ./ModelfileLlama2Jetson ollama run llama2-jetson ``` and ``` dillera@jetson:~$ cat ./ModelfileLlama2Jetson FROM llama2 PARAMETER num_gpu 999 ``` thanks!
GiteaMirror commented 2026-04-12 10:21:13 -05:00
Author
Owner
Copy Link

@bnodnarb commented on GitHub (Nov 28, 2023):

Hey @dillera - Thanks for sharing those logs. Llama2 7B should fit on the 8GB Jetson. Even if it didn't fit, JTOP should be pumping out increased GPU usage on the GPU tab.

I think @mraiser was able to get it to work with the instructions (please see here: https://github.com/jmorganca/ollama/issues/1071#issuecomment-1807134437) , though I'm not sure which model they were running.

If you have a fresh microSD card laying around, I might try it on a fresh install.

I'm sure you're doing it correctly, nonetheless I will update the tutorial with a single copy & paste snippet of commands that will get it running on a fresh install of Jetpack 5.1.2. I'll tailor it for Llama2 as well so it will fit your specific use case.

I'll post here when I've submitted the pull request.

Thanks!

<!-- gh-comment-id:1829259622 --> @bnodnarb commented on GitHub (Nov 28, 2023): Hey @dillera - Thanks for sharing those logs. Llama2 7B should fit on the 8GB Jetson. Even if it didn't fit, JTOP should be pumping out increased GPU usage on the GPU tab. I think @mraiser was able to get it to work with the instructions (please see here: https://github.com/jmorganca/ollama/issues/1071#issuecomment-1807134437) , though I'm not sure which model they were running. If you have a fresh microSD card laying around, I might try it on a fresh install. I'm sure you're doing it correctly, nonetheless I will update the tutorial with a single copy & paste snippet of commands that will get it running on a fresh install of Jetpack 5.1.2. I'll tailor it for Llama2 as well so it will fit your specific use case. I'll post here when I've submitted the pull request. Thanks!
GiteaMirror commented 2026-04-12 10:21:13 -05:00
Author
Owner
Copy Link

@bnodnarb commented on GitHub (Nov 28, 2023):

Hi @dillera - If you have a spare Micro SD card, I've added a quickstart guide that will get you up and running with a single copy/paste command - please see Pull Request here: https://github.com/jmorganca/ollama/pull/1297

This includes a quickstart guide still uses Mistral to stay consistent with the rest of the tutorial, but I think it would work perfectly with Llama2 as well (I will test that part).

<!-- gh-comment-id:1829487947 --> @bnodnarb commented on GitHub (Nov 28, 2023): Hi @dillera - If you have a spare Micro SD card, I've added a quickstart guide that will get you up and running with a single copy/paste command - please see Pull Request here: https://github.com/jmorganca/ollama/pull/1297 This includes a quickstart guide still uses Mistral to stay consistent with the rest of the tutorial, but I think it would work perfectly with Llama2 as well (I will test that part).
GiteaMirror commented 2026-04-12 10:21:13 -05:00
Author
Owner
Copy Link

@bnodnarb commented on GitHub (Nov 28, 2023):

Interesting discovery, it appears that Llama2 might require more RAM than is available on the 8GB Jetson. I can run Mistral and other models, but the Ollama default Llama2 model appears to be capping out.

<!-- gh-comment-id:1829531828 --> @bnodnarb commented on GitHub (Nov 28, 2023): Interesting discovery, it appears that Llama2 might require more RAM than is available on the 8GB Jetson. I can run Mistral and other models, but the Ollama default Llama2 model appears to be capping out.
GiteaMirror commented 2026-04-12 10:21:14 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

Ok so it's just the model and RAM? But how does it run on the CPUs? It's just not touching the GPU...

In fact, I have an ancient HP GL3xxx Xeon server that ends up being faster than the Jetson (albiet with only CPUS)...

HP:

Once upon a time, in the magical kingdom of Azura, there lived a little girl named Lily. geprüft. She had long, curly blonde hair and big blue eyes that twinkled with mischief. Lily was a curious child who loved to explore the world around her, discovering new sights and sounds at every turn.
One day, while wandering through the forest near her home, Lily stumbled upon a hidden glade. In the center of the glade stood an enormous tree unlike any she had ever seen before. Its trunk was covered in moss and vines, and its branches reached
llama_print_timings:        load time =     613.45 ms
llama_print_timings:      sample time =      78.14 ms /   128 runs   (    0.61 ms per token,  1638.06 tokens per second)
llama_print_timings: prompt eval time =     734.53 ms /     6 tokens (  122.42 ms per token,     8.17 tokens per second)
llama_print_timings:        eval time =   37202.16 ms /   127 runs   (  292.93 ms per token,     3.41 tokens per second)
llama_print_timings:       total time =   38236.38 ms
Log end

Jetston Orin Nano 8GB:

 Once upon a time, in the land of the Fjords,
 Unterscheidung between men and women was not a cultural phenomenon.  The people lived in harmony with nature and were governed by their own laws and customs. They believed that all living things were interconnected and that each person had a special purpose in life.
One day, a young man named Eirik set out on a journey to explore the world beyond his village. He traveled through mountains and forests, over rivers and across the sea, until he came to a great city where he met a wise old man who taught him about the importance of understanding oneself
llama_print_timings:        load time =  1612.25 ms
llama_print_timings:      sample time =   125.32 ms /   128 runs   (    0.98 ms per token,  1021.36 tokens per second)
llama_print_timings: prompt eval time =  2571.50 ms /     6 tokens (  428.58 ms per token,     2.33 tokens per second)
llama_print_timings:        eval time = 58262.27 ms /   127 runs   (  458.76 ms per token,     2.18 tokens per second)
llama_print_timings:       total time = 61062.66 ms
Log end

I'll take a look at 1297 and see. I'm pretty sure my install of the os and jetpack is good, I've installed and used many many linux systems before this and it was just an image flashed to a 32M SD card - but then I copied it to a 1TB nvme and am using that as root.

<!-- gh-comment-id:1829886621 --> @dillera commented on GitHub (Nov 28, 2023): Ok so it's just the model and RAM? But how does it run on the CPUs? It's just not touching the GPU... In fact, I have an ancient HP GL3xxx Xeon server that ends up being faster than the Jetson (albiet with only CPUS)... HP: ``` Once upon a time, in the magical kingdom of Azura, there lived a little girl named Lily. geprüft. She had long, curly blonde hair and big blue eyes that twinkled with mischief. Lily was a curious child who loved to explore the world around her, discovering new sights and sounds at every turn. One day, while wandering through the forest near her home, Lily stumbled upon a hidden glade. In the center of the glade stood an enormous tree unlike any she had ever seen before. Its trunk was covered in moss and vines, and its branches reached llama_print_timings: load time = 613.45 ms llama_print_timings: sample time = 78.14 ms / 128 runs ( 0.61 ms per token, 1638.06 tokens per second) llama_print_timings: prompt eval time = 734.53 ms / 6 tokens ( 122.42 ms per token, 8.17 tokens per second) llama_print_timings: eval time = 37202.16 ms / 127 runs ( 292.93 ms per token, 3.41 tokens per second) llama_print_timings: total time = 38236.38 ms Log end ``` Jetston Orin Nano 8GB: ``` Once upon a time, in the land of the Fjords, Unterscheidung between men and women was not a cultural phenomenon. The people lived in harmony with nature and were governed by their own laws and customs. They believed that all living things were interconnected and that each person had a special purpose in life. One day, a young man named Eirik set out on a journey to explore the world beyond his village. He traveled through mountains and forests, over rivers and across the sea, until he came to a great city where he met a wise old man who taught him about the importance of understanding oneself llama_print_timings: load time = 1612.25 ms llama_print_timings: sample time = 125.32 ms / 128 runs ( 0.98 ms per token, 1021.36 tokens per second) llama_print_timings: prompt eval time = 2571.50 ms / 6 tokens ( 428.58 ms per token, 2.33 tokens per second) llama_print_timings: eval time = 58262.27 ms / 127 runs ( 458.76 ms per token, 2.18 tokens per second) llama_print_timings: total time = 61062.66 ms Log end ``` I'll take a look at 1297 and see. I'm pretty sure my install of the os and jetpack is good, I've installed and used many many linux systems before this and it was just an image flashed to a 32M SD card - but then I copied it to a 1TB nvme and am using that as root.
GiteaMirror commented 2026-04-12 10:21:14 -05:00
Author
Owner
Copy Link

@mraiser commented on GitHub (Nov 28, 2023):

I think @mraiser was able to get it to work with the instructions (please see here: #1071 (comment)) , though I'm not sure which model they were running.

I had used mistral. I am able to get llama2 to load by reducing the num_gpu parameter to 20. Leaving it at 999 froze the system when trying to load llama2. At 20 it will load, but the entire system slows to a crawl, and even at 5 performance was worse than cpu-only. That's as far as I got with it before switching back to mistral.

<!-- gh-comment-id:1829919051 --> @mraiser commented on GitHub (Nov 28, 2023): > I think @mraiser was able to get it to work with the instructions (please see here: [#1071 (comment)](https://github.com/jmorganca/ollama/issues/1071#issuecomment-1807134437)) , though I'm not sure which model they were running. I had used mistral. I am able to get llama2 to load by reducing the num_gpu parameter to 20. Leaving it at 999 froze the system when trying to load llama2. At 20 it will load, but the entire system slows to a crawl, and even at 5 performance was worse than cpu-only. That's as far as I got with it before switching back to mistral.
GiteaMirror commented 2026-04-12 10:21:15 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

Interesting. I will play with num_gps and mistral then- I just want to see this GPU active, or it's no better than the old xeon server I have sitting here!

I have also had the Jetson lock up on me while returning a result with the llama2 model loaded.

I've also played with the docker image using llama.cpp and it also will not use the GPU...
https://github.com/dusty-nv/jetson-containers/tree/master/packages/llm/llama_cpp

<!-- gh-comment-id:1830051626 --> @dillera commented on GitHub (Nov 28, 2023): Interesting. I will play with num_gps and mistral then- I just want to see this GPU active, or it's no better than the old xeon server I have sitting here! I have also had the Jetson lock up on me while returning a result with the llama2 model loaded. I've also played with the docker image using llama.cpp and it also will not use the GPU... https://github.com/dusty-nv/jetson-containers/tree/master/packages/llm/llama_cpp
GiteaMirror commented 2026-04-12 10:21:15 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

Another thing I'm finding: the "8GB" Jetson Nano really has only 6.3GB ram available.

The rest is being used for some sort of RAM/SWAP and can only be fixed with a re-flash of the QSPI... Disabling swap does not free up this ram, you have to flash it.

https://jetsonhacks.com/2023/05/26/jetson-orin-nano-flashing-qspi-firmware-for-more-memory/

a really good command I found is:

~$ sudo tegrastats
11-28-2023 10:30:16 RAM 1175/6481MB (lfb 1148x4MB) CPU [3%@729,1%@729,2%@729,1%@729,0%@729,2%@729] EMC_FREQ 0%@2133 GR3D_FREQ 0%@[0,305] VIC_FREQ 435 APE 200 CV0@-256C CPU@46.781C SOC2@45.937C SOC0@44.562C CV1@-256C GPU@45.75C tj@46.781C SOC

Notice the RAM 1175/6481MB - there is only 6.4GB of the 8 available for Linux...

So these 8GB models will not fit.

Another good command:

$ sudo lshw -C memory
<!-- gh-comment-id:1830149671 --> @dillera commented on GitHub (Nov 28, 2023): Another thing I'm finding: the "8GB" Jetson Nano really has only 6.3GB ram available. The rest is being used for some sort of RAM/SWAP and can only be fixed with a re-flash of the QSPI... Disabling swap does not free up this ram, you have to flash it. https://jetsonhacks.com/2023/05/26/jetson-orin-nano-flashing-qspi-firmware-for-more-memory/ a really good command I found is: ``` ~$ sudo tegrastats 11-28-2023 10:30:16 RAM 1175/6481MB (lfb 1148x4MB) CPU [3%@729,1%@729,2%@729,1%@729,0%@729,2%@729] EMC_FREQ 0%@2133 GR3D_FREQ 0%@[0,305] VIC_FREQ 435 APE 200 CV0@-256C CPU@46.781C SOC2@45.937C SOC0@44.562C CV1@-256C GPU@45.75C tj@46.781C SOC ``` Notice the RAM 1175/6481MB - there is only 6.4GB of the 8 available for Linux... So these 8GB models will not fit. Another good command: ``` $ sudo lshw -C memory ```
GiteaMirror commented 2026-04-12 10:21:15 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

also:

echo 'Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64"' | sudo tee -a /etc/systemd/system/ollama.service

Is a lot cleaner than that tee command, and permanent.

<!-- gh-comment-id:1830169121 --> @dillera commented on GitHub (Nov 28, 2023): also: ``` echo 'Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64"' | sudo tee -a /etc/systemd/system/ollama.service ``` Is a lot cleaner than that tee command, and permanent.
GiteaMirror commented 2026-04-12 10:21:15 -05:00
Author
Owner
Copy Link

@mraiser commented on GitHub (Nov 28, 2023):

Another thing I'm finding: the "8GB" Jetson Nano really has only 6.3GB ram available.

The rest is being used for some sort of RAM/SWAP and can only be fixed with a re-flash of the QSPI... Disabling swap does not free up this ram, you have to flash it.

https://jetsonhacks.com/2023/05/26/jetson-orin-nano-flashing-qspi-firmware-for-more-memory/

That process frees up almost 1gb-- definitely worth it.

<!-- gh-comment-id:1830178013 --> @mraiser commented on GitHub (Nov 28, 2023): > Another thing I'm finding: the "8GB" Jetson Nano really has only 6.3GB ram available. > > The rest is being used for some sort of RAM/SWAP and can only be fixed with a re-flash of the QSPI... Disabling swap does not free up this ram, you have to flash it. > > https://jetsonhacks.com/2023/05/26/jetson-orin-nano-flashing-qspi-firmware-for-more-memory/ That process frees up almost 1gb-- definitely worth it.
GiteaMirror commented 2026-04-12 10:21:16 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

I cannot get mistral or llama2 to touch the gpu. It's not RAM- if it were ram the model wouldn't load. It loads and I can ask it prompts. I only see CPU util. on jtop. GPU is never touched.

Starting ollmama:

Nov 28 12:44:17 jetson ollama[55387]: 2023/11/28 12:44:17 routes.go:797: warning: gpu support may not be enabled, check that you have installed GPU drivers: nvidia-smi command failed
Nov 28 12:46:06 jetson ollama[55387]: [GIN] 2023/11/28 - 12:46:06 | 200 |      66.595µs |       127.0.0.1 | HEAD     "/"
Nov 28 12:46:06 jetson ollama[55387]: [GIN] 2023/11/28 - 12:46:06 | 200 |     791.978µs |       127.0.0.1 | POST     "/api/show"
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:420: starting llama runner
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:478: waiting for llama runner to start responding
Nov 28 12:46:07 jetson ollama[55387]: NvRmMemInitNvmap failed with Permission denied
Nov 28 12:46:07 jetson ollama[55387]: 549: Memory Manager Not supported
Nov 28 12:46:07 jetson ollama[55387]: ****NvRmMemInit failed**** error type: 196626
Nov 28 12:46:07 jetson ollama[55387]: *** NvRmMemInit failed NvRmMemConstructor
Nov 28 12:46:07 jetson ollama[55387]: CUDA error 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported
Nov 28 12:46:07 jetson ollama[55387]: current device: 0
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:435: 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported
Nov 28 12:46:07 jetson ollama[55387]: current device: 0
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:443: error starting llama runner: llama runner process has terminated
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:509: llama runner stopped successfully
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:420: starting llama runner
Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:478: waiting for llama runner to start responding
Nov 28 12:46:07 jetson ollama[56294]: {"timestamp":1701193567,"level":"WARNING","function":"server_params_parse","line":873,"message":"Not compiled with GPU offload support, --n-gpu-layers option will be ignored. See main README.md for information on enabling GPU BLAS support","n_gpu_layers":-1}

Again, this is using the env variable set in the systemd service:

[Unit]
Description=Ollama Service
After=network-online.target

[Service]
ExecStart=/usr/local/bin/ollama serve
User=ollama
Group=ollama
Restart=always
RestartSec=3
Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64"
Environment="PATH=/usr/local/cuda/lib64:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin"

[Install]
WantedBy=default.target

the Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64" like you do in your tee.

@mraiser - is to too much to ask to see some sort of log from ollama from mistral that shows it hitting, using the GPU in any way? i.e. output from:

journalctl -fxeu ollama.service

I would greatly appreciate it. Otherwise, i'm done w/ this, might as well just focus on using a mac w/ llama.ccp.

<!-- gh-comment-id:1830383699 --> @dillera commented on GitHub (Nov 28, 2023): I cannot get mistral or llama2 to touch the gpu. It's not RAM- if it were ram the model wouldn't load. It loads and I can ask it prompts. I only see CPU util. on jtop. GPU is never touched. Starting ollmama: ``` Nov 28 12:44:17 jetson ollama[55387]: 2023/11/28 12:44:17 routes.go:797: warning: gpu support may not be enabled, check that you have installed GPU drivers: nvidia-smi command failed Nov 28 12:46:06 jetson ollama[55387]: [GIN] 2023/11/28 - 12:46:06 | 200 | 66.595µs | 127.0.0.1 | HEAD "/" Nov 28 12:46:06 jetson ollama[55387]: [GIN] 2023/11/28 - 12:46:06 | 200 | 791.978µs | 127.0.0.1 | POST "/api/show" Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:420: starting llama runner Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:478: waiting for llama runner to start responding Nov 28 12:46:07 jetson ollama[55387]: NvRmMemInitNvmap failed with Permission denied Nov 28 12:46:07 jetson ollama[55387]: 549: Memory Manager Not supported Nov 28 12:46:07 jetson ollama[55387]: ****NvRmMemInit failed**** error type: 196626 Nov 28 12:46:07 jetson ollama[55387]: *** NvRmMemInit failed NvRmMemConstructor Nov 28 12:46:07 jetson ollama[55387]: CUDA error 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported Nov 28 12:46:07 jetson ollama[55387]: current device: 0 Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:435: 801 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/gguf/ggml-cuda.cu:5661: operation not supported Nov 28 12:46:07 jetson ollama[55387]: current device: 0 Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:443: error starting llama runner: llama runner process has terminated Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:509: llama runner stopped successfully Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:420: starting llama runner Nov 28 12:46:07 jetson ollama[55387]: 2023/11/28 12:46:07 llama.go:478: waiting for llama runner to start responding Nov 28 12:46:07 jetson ollama[56294]: {"timestamp":1701193567,"level":"WARNING","function":"server_params_parse","line":873,"message":"Not compiled with GPU offload support, --n-gpu-layers option will be ignored. See main README.md for information on enabling GPU BLAS support","n_gpu_layers":-1} ``` Again, this is using the env variable set in the systemd service: ``` [Unit] Description=Ollama Service After=network-online.target [Service] ExecStart=/usr/local/bin/ollama serve User=ollama Group=ollama Restart=always RestartSec=3 Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64" Environment="PATH=/usr/local/cuda/lib64:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin" [Install] WantedBy=default.target ``` the Environment="LD_LIBRARY_PATH=/usr/local/cuda/lib64" like you do in your tee. @mraiser - is to too much to ask to see some sort of log from ollama from mistral that shows it hitting, using the GPU in any way? i.e. output from: journalctl -fxeu ollama.service I would greatly appreciate it. Otherwise, i'm done w/ this, might as well just focus on using a mac w/ llama.ccp.
GiteaMirror commented 2026-04-12 10:21:16 -05:00
Author
Owner
Copy Link

@mraiser commented on GitHub (Nov 28, 2023):

@mraiser - is to too much to ask to see some sort of log from ollama from mistral that shows it hitting, using the GPU in any way? i.e. output from:

Looks like you have some kind of permissions issue, like the ollama user can't access your nvidia drivers or something. Maybe related to running as a service instead of from command line?

In any case, here's my output from ollama showing mistral running:

mraiser@jon:~$ LD_LIBRARY_PATH=/usr/local/cuda/lib64 ollama serve
2023/11/28 13:49:20 images.go:824: total blobs: 19
2023/11/28 13:49:20 images.go:831: total unused blobs removed: 0
2023/11/28 13:49:20 routes.go:696: Listening on 127.0.0.1:11434 (version 0.1.9)
2023/11/28 13:49:20 routes.go:716: Warning: GPU support may not be enabled, check you have installed GPU drivers: nvidia-smi command failed
[GIN] 2023/11/28 - 13:49:33 | 200 | 136.52µs | 127.0.0.1 | HEAD "/"
[GIN] 2023/11/28 - 13:49:33 | 200 | 140.815582ms | 127.0.0.1 | POST "/api/show"
2023/11/28 13:49:39 llama.go:415: starting llama runner
2023/11/28 13:49:39 llama.go:473: waiting for llama runner to start responding
ggml_init_cublas: found 1 CUDA devices:
Device 0: Orin, compute capability 8.7
{"timestamp":1701197397,"level":"INFO","function":"main","line":1324,"message":"build info","build":219,"commit":"9e70cc0"}
{"timestamp":1701197397,"level":"INFO","function":"main","line":1330,"message":"system info","n_threads":3,"n_threads_batch":-1,"total_threads":6,"system_info":"AVX = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | SSSE3 = 0 | VSX = 0 | "}
llama_model_loader: loaded meta data with 20 key-value pairs and 291 tensors from /home/mraiser/.ollama/models/blobs/sha256:6ae28029995007a3ee8d0b8556d50f3b59b831074cf19c84de87acf51fb54054 (version GGUF V2 (latest))
llama_model_loader: - tensor 0: token_embd.weight q4_0 [ 4096, 32000, 1, 1 ]
llama_model_loader: - tensor 1: blk.0.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ]
llama_model_loader: - tensor 2: blk.0.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ]
llama_model_loader: - tensor 3: blk.0.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ]

[... snip ...]

llm_load_print_meta: model ftype = mostly Q4_0
llm_load_print_meta: model params = 7.24 B
llm_load_print_meta: model size = 3.83 GiB (4.54 BPW)
llm_load_print_meta: general.name = mistralai
llm_load_print_meta: BOS token = 1 ''
llm_load_print_meta: EOS token = 2 ''
llm_load_print_meta: UNK token = 0 ''
llm_load_print_meta: LF token = 13 '<0x0A>'
llm_load_tensors: ggml ctx size = 0.10 MB
llm_load_tensors: using CUDA for GPU acceleration
llm_load_tensors: mem required = 70.41 MB
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 35/35 layers to GPU
llm_load_tensors: VRAM used: 3847.55 MB
..................................................................................................
llama_new_context_with_model: n_ctx = 2048
llama_new_context_with_model: freq_base = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: offloading v cache to GPU
llama_kv_cache_init: offloading k cache to GPU
llama_kv_cache_init: VRAM kv self = 256.00 MB
llama_new_context_with_model: kv self size = 256.00 MB
llama_new_context_with_model: compute buffer total size = 162.13 MB
llama_new_context_with_model: VRAM scratch buffer: 156.00 MB
llama_new_context_with_model: total VRAM used: 4259.56 MB (model: 3847.55 MB, context: 412.00 MB)

llama server listening at http://127.0.0.1:53649

{"timestamp":1701197468,"level":"INFO","function":"main","line":1749,"message":"HTTP server listening","hostname":"127.0.0.1","port":53649}
{"timestamp":1701197468,"level":"INFO","function":"log_server_request","line":1240,"message":"request","remote_addr":"127.0.0.1","remote_port":44018,"status":200,"method":"HEAD","path":"/","params":{}}
2023/11/28 13:51:08 llama.go:487: llama runner started in 89.449249 seconds
[GIN] 2023/11/28 - 13:51:08 | 200 | 1m35s | 127.0.0.1 | POST "/api/generate"

<!-- gh-comment-id:1830487014 --> @mraiser commented on GitHub (Nov 28, 2023): > @mraiser - is to too much to ask to see some sort of log from ollama from mistral that shows it hitting, using the GPU in any way? i.e. output from: Looks like you have some kind of permissions issue, like the ollama user can't access your nvidia drivers or something. Maybe related to running as a service instead of from command line? In any case, here's my output from ollama showing mistral running: mraiser@jon:~$ LD_LIBRARY_PATH=/usr/local/cuda/lib64 ollama serve 2023/11/28 13:49:20 images.go:824: total blobs: 19 2023/11/28 13:49:20 images.go:831: total unused blobs removed: 0 2023/11/28 13:49:20 routes.go:696: Listening on 127.0.0.1:11434 (version 0.1.9) 2023/11/28 13:49:20 routes.go:716: Warning: GPU support may not be enabled, check you have installed GPU drivers: nvidia-smi command failed [GIN] 2023/11/28 - 13:49:33 | 200 | 136.52µs | 127.0.0.1 | HEAD "/" [GIN] 2023/11/28 - 13:49:33 | 200 | 140.815582ms | 127.0.0.1 | POST "/api/show" 2023/11/28 13:49:39 llama.go:415: starting llama runner 2023/11/28 13:49:39 llama.go:473: waiting for llama runner to start responding ggml_init_cublas: found 1 CUDA devices: Device 0: Orin, compute capability 8.7 {"timestamp":1701197397,"level":"INFO","function":"main","line":1324,"message":"build info","build":219,"commit":"9e70cc0"} {"timestamp":1701197397,"level":"INFO","function":"main","line":1330,"message":"system info","n_threads":3,"n_threads_batch":-1,"total_threads":6,"system_info":"AVX = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 0 | NEON = 1 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | SSSE3 = 0 | VSX = 0 | "} llama_model_loader: loaded meta data with 20 key-value pairs and 291 tensors from /home/mraiser/.ollama/models/blobs/sha256:6ae28029995007a3ee8d0b8556d50f3b59b831074cf19c84de87acf51fb54054 (version GGUF V2 (latest)) llama_model_loader: - tensor 0: token_embd.weight q4_0 [ 4096, 32000, 1, 1 ] llama_model_loader: - tensor 1: blk.0.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 2: blk.0.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 3: blk.0.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] [... snip ...] llm_load_print_meta: model ftype = mostly Q4_0 llm_load_print_meta: model params = 7.24 B llm_load_print_meta: model size = 3.83 GiB (4.54 BPW) llm_load_print_meta: general.name = mistralai llm_load_print_meta: BOS token = 1 '<s>' llm_load_print_meta: EOS token = 2 '</s>' llm_load_print_meta: UNK token = 0 '<unk>' llm_load_print_meta: LF token = 13 '<0x0A>' llm_load_tensors: ggml ctx size = 0.10 MB llm_load_tensors: using CUDA for GPU acceleration llm_load_tensors: mem required = 70.41 MB llm_load_tensors: offloading 32 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 35/35 layers to GPU llm_load_tensors: VRAM used: 3847.55 MB .................................................................................................. llama_new_context_with_model: n_ctx = 2048 llama_new_context_with_model: freq_base = 10000.0 llama_new_context_with_model: freq_scale = 1 llama_kv_cache_init: offloading v cache to GPU llama_kv_cache_init: offloading k cache to GPU llama_kv_cache_init: VRAM kv self = 256.00 MB llama_new_context_with_model: kv self size = 256.00 MB llama_new_context_with_model: compute buffer total size = 162.13 MB llama_new_context_with_model: VRAM scratch buffer: 156.00 MB llama_new_context_with_model: total VRAM used: 4259.56 MB (model: 3847.55 MB, context: 412.00 MB) llama server listening at http://127.0.0.1:53649 {"timestamp":1701197468,"level":"INFO","function":"main","line":1749,"message":"HTTP server listening","hostname":"127.0.0.1","port":53649} {"timestamp":1701197468,"level":"INFO","function":"log_server_request","line":1240,"message":"request","remote_addr":"127.0.0.1","remote_port":44018,"status":200,"method":"HEAD","path":"/","params":{}} 2023/11/28 13:51:08 llama.go:487: llama runner started in 89.449249 seconds [GIN] 2023/11/28 - 13:51:08 | 200 | 1m35s | 127.0.0.1 | POST "/api/generate"
GiteaMirror commented 2026-04-12 10:21:16 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

@mraiser thanks! just what I was looking for!

To be perfectly clear, you are running ollama from the command line (as your user, yes I see that now), then in another window doing something like:

$ ollama pull mistral
$ ollama create mistral-jetson -f ./Modelfile_mistral_jetson
$ ollama run mistral-jetson
[prompt]

EDIT:

It works! Thanks for this tip.

my own logs show it:

2023/11/28 14:34:37 llama.go:420: starting llama runner
2023/11/28 14:34:37 llama.go:478: waiting for llama runner to start responding
ggml_init_cublas: found 1 CUDA devices:
  Device 0: Orin, compute capability 8.7

and I see the GPU moving on jtop finally.

<!-- gh-comment-id:1830541504 --> @dillera commented on GitHub (Nov 28, 2023): @mraiser thanks! just what I was looking for! To be perfectly clear, you are running ollama from the command line (as your user, yes I see that now), then in another window doing something like: ``` $ ollama pull mistral $ ollama create mistral-jetson -f ./Modelfile_mistral_jetson $ ollama run mistral-jetson [prompt] ``` EDIT: It works! Thanks for this tip. my own logs show it: ``` 2023/11/28 14:34:37 llama.go:420: starting llama runner 2023/11/28 14:34:37 llama.go:478: waiting for llama runner to start responding ggml_init_cublas: found 1 CUDA devices: Device 0: Orin, compute capability 8.7 ``` and I see the GPU moving on jtop finally.
GiteaMirror commented 2026-04-12 10:21:17 -05:00
Author
Owner
Copy Link

@dillera commented on GitHub (Nov 28, 2023):

Need to run ollama service with proper permssions, at very least launch as default user on Jetson Nano.

ensure you see (after running a model with ollama run in another window):

dillera@jetson:~$ LD_LIBRARY_PATH=/usr/local/cuda/lib64 ollama serve
[...]
2023/11/28 14:34:37 llama.go:478: waiting for llama runner to start responding
ggml_init_cublas: found 1 CUDA devices:
  Device 0: Orin, compute capability 8.7
[...]

and at start and after a prompt:

llm_load_tensors: using CUDA for GPU acceleration
llm_load_tensors: mem required  = 3683.90 MB
llm_load_tensors: offloading 2 repeating layers to GPU
llm_load_tensors: offloaded 2/35 layers to GPU

in the output; other nvidia-smi commands failing at the start of ollama is normal for the Jetson.

<!-- gh-comment-id:1830552657 --> @dillera commented on GitHub (Nov 28, 2023): Need to run ollama service with proper permssions, at very least launch as default user on Jetson Nano. ensure you see (after running a model with ollama run in another window): ``` dillera@jetson:~$ LD_LIBRARY_PATH=/usr/local/cuda/lib64 ollama serve [...] 2023/11/28 14:34:37 llama.go:478: waiting for llama runner to start responding ggml_init_cublas: found 1 CUDA devices: Device 0: Orin, compute capability 8.7 [...] ``` and at start and after a prompt: ``` llm_load_tensors: using CUDA for GPU acceleration llm_load_tensors: mem required = 3683.90 MB llm_load_tensors: offloading 2 repeating layers to GPU llm_load_tensors: offloaded 2/35 layers to GPU ``` in the output; other `nvidia-smi commands` failing at the start of ollama is normal for the Jetson.
GiteaMirror commented 2026-04-12 10:21:17 -05:00
Author
Owner
Copy Link

@Q-point commented on GitHub (Jan 20, 2024):

I am following the same instructions here and I don't see it using the GPU.

dhq@dhq-desktop:~$ LD_LIBRARY_PATH=/usr/local/cuda/lib64 ollama serve
2024/01/19 21:25:40 images.go:808: total blobs: 8
2024/01/19 21:25:40 images.go:815: total unused blobs removed: 0
2024/01/19 21:25:40 routes.go:930: Listening on 127.0.0.1:11434 (version 0.1.20)
2024/01/19 21:25:41 shim_ext_server.go:142: Dynamic LLM variants [cuda]
2024/01/19 21:25:41 gpu.go:88: Detecting GPU type
2024/01/19 21:25:41 gpu.go:203: Searching for GPU management library libnvidia-ml.so
2024/01/19 21:25:41 gpu.go:248: Discovered GPU libraries: []
2024/01/19 21:25:41 gpu.go:203: Searching for GPU management library librocm_smi64.so
2024/01/19 21:25:41 gpu.go:248: Discovered GPU libraries: []
2024/01/19 21:25:41 routes.go:953: no GPU detected

[GIN] 2024/01/19 - 21:26:07 | 200 |     151.903µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:26:07 | 404 |     493.628µs |       127.0.0.1 | POST     "/api/show"
[GIN] 2024/01/19 - 21:26:07 | 200 |  345.212605ms |       127.0.0.1 | POST     "/api/pull"
[GIN] 2024/01/19 - 21:26:26 | 200 |      76.704µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:26:26 | 404 |     233.598µs |       127.0.0.1 | POST     "/api/show"
[GIN] 2024/01/19 - 21:26:26 | 200 |   216.31751ms |       127.0.0.1 | POST     "/api/pull"
[GIN] 2024/01/19 - 21:26:42 | 200 |      72.608µs |       127.0.0.1 | HEAD     "/"
2024/01/19 21:26:43 images.go:430: [model] - mistral
2024/01/19 21:26:44 download.go:123: downloading e8a35b5937a5 in 42 100 MB part(s)
[GIN] 2024/01/19 - 21:31:18 | 200 |       88.32µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:31:18 | 200 |   14.408573ms |       127.0.0.1 | GET      "/api/tags"
[GIN] 2024/01/19 - 21:31:42 | 200 |     102.623µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:31:42 | 404 |     444.319µs |       127.0.0.1 | DELETE   "/api/delete"
[GIN] 2024/01/19 - 21:31:45 | 200 |      78.719µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:31:45 | 200 |    1.528636ms |       127.0.0.1 | DELETE   "/api/delete"
[GIN] 2024/01/19 - 21:31:50 | 200 |      87.455µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:31:50 | 200 |      880.67µs |       127.0.0.1 | GET      "/api/tags"
2024/01/19 21:38:28 download.go:162: e8a35b5937a5 part 10 attempt 0 failed: unexpected EOF, retrying in 1s
2024/01/19 21:39:20 download.go:123: downloading 43070e2d4e53 in 1 11 KB part(s)
2024/01/19 21:39:23 download.go:123: downloading e6836092461f in 1 42 B part(s)
2024/01/19 21:39:26 download.go:123: downloading ed11eda7790d in 1 30 B part(s)
2024/01/19 21:39:28 download.go:123: downloading f9b1e3196ecf in 1 483 B part(s)
2024/01/19 21:39:36 images.go:430: [num_gpu] - 100
[GIN] 2024/01/19 - 21:39:36 | 200 |        12m53s |       127.0.0.1 | POST     "/api/create"
[GIN] 2024/01/19 - 21:40:10 | 200 |      85.567µs |       127.0.0.1 | HEAD     "/"
[GIN] 2024/01/19 - 21:40:10 | 200 |    2.897401ms |       127.0.0.1 | POST     "/api/show"
[GIN] 2024/01/19 - 21:40:10 | 200 |    1.145758ms |       127.0.0.1 | POST     "/api/show"
2024/01/19 21:40:10 llm.go:71: GPU not available, falling back to CPU
2024/01/19 21:40:10 ext_server_common.go:136: Initializing internal llama server
llama_model_loader: loaded meta data with 24 key-value pairs and 291 tensors from /home/dhq/.ollama/models/blobs/sha256:e8a35b5937a5e6d5c35d1f2a15f161e07eefe5e5bb0a3cdd42998ee79b057730 (version GGUF V3 (latest))
llama_model_loader: - tensor    0:                token_embd.weight q4_0     [  4096, 32000,     1,     1 ]
llama_model_loader: - tensor    1:           blk.0.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor    2:            blk.0.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor    3:            blk.0.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor    4:              blk.0.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor    5:            blk.0.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor    6:              blk.0.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor    7:         blk.0.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor    8:              blk.0.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor    9:              blk.0.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   10:           blk.1.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   11:            blk.1.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   12:            blk.1.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   13:              blk.1.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   14:            blk.1.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   15:              blk.1.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   16:         blk.1.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   17:              blk.1.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   18:              blk.1.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   19:           blk.10.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   20:             blk.10.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   21:             blk.10.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   22:        blk.10.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   23:             blk.10.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   24:             blk.10.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   25:           blk.2.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   26:            blk.2.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   27:            blk.2.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   28:              blk.2.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   29:            blk.2.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   30:              blk.2.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   31:         blk.2.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   32:              blk.2.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   33:              blk.2.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   34:           blk.3.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   35:            blk.3.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   36:            blk.3.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   37:              blk.3.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   38:            blk.3.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   39:              blk.3.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   40:         blk.3.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   41:              blk.3.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   42:              blk.3.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   43:           blk.4.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   44:            blk.4.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   45:            blk.4.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   46:              blk.4.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   47:            blk.4.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   48:              blk.4.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   49:         blk.4.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   50:              blk.4.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   51:              blk.4.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   52:           blk.5.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   53:            blk.5.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   54:            blk.5.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   55:              blk.5.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   56:            blk.5.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   57:              blk.5.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   58:         blk.5.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   59:              blk.5.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   60:              blk.5.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   61:           blk.6.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   62:            blk.6.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   63:            blk.6.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   64:              blk.6.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   65:            blk.6.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   66:              blk.6.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   67:         blk.6.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   68:              blk.6.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   69:              blk.6.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   70:           blk.7.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   71:            blk.7.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   72:            blk.7.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   73:              blk.7.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   74:            blk.7.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   75:              blk.7.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   76:         blk.7.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   77:              blk.7.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   78:              blk.7.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   79:           blk.8.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   80:            blk.8.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   81:            blk.8.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   82:              blk.8.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   83:            blk.8.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   84:              blk.8.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   85:         blk.8.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   86:              blk.8.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   87:              blk.8.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   88:           blk.9.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   89:            blk.9.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   90:            blk.9.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   91:              blk.9.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor   92:            blk.9.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   93:              blk.9.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   94:         blk.9.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   95:              blk.9.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor   96:              blk.9.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor   97:          blk.10.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor   98:           blk.10.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor   99:           blk.10.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  100:          blk.11.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  101:           blk.11.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  102:           blk.11.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  103:             blk.11.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  104:           blk.11.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  105:             blk.11.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  106:        blk.11.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  107:             blk.11.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  108:             blk.11.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  109:          blk.12.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  110:           blk.12.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  111:           blk.12.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  112:             blk.12.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  113:           blk.12.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  114:             blk.12.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  115:        blk.12.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  116:             blk.12.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  117:             blk.12.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  118:          blk.13.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  119:           blk.13.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  120:           blk.13.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  121:             blk.13.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  122:           blk.13.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  123:             blk.13.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  124:        blk.13.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  125:             blk.13.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  126:             blk.13.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  127:          blk.14.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  128:           blk.14.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  129:           blk.14.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  130:             blk.14.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  131:           blk.14.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  132:             blk.14.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  133:        blk.14.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  134:             blk.14.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  135:             blk.14.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  136:          blk.15.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  137:           blk.15.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  138:           blk.15.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  139:             blk.15.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  140:           blk.15.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  141:             blk.15.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  142:        blk.15.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  143:             blk.15.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  144:             blk.15.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  145:          blk.16.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  146:           blk.16.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  147:           blk.16.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  148:             blk.16.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  149:           blk.16.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  150:             blk.16.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  151:        blk.16.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  152:             blk.16.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  153:             blk.16.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  154:          blk.17.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  155:           blk.17.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  156:           blk.17.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  157:             blk.17.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  158:           blk.17.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  159:             blk.17.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  160:        blk.17.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  161:             blk.17.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  162:             blk.17.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  163:          blk.18.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  164:           blk.18.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  165:           blk.18.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  166:             blk.18.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  167:           blk.18.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  168:             blk.18.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  169:        blk.18.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  170:             blk.18.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  171:             blk.18.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  172:          blk.19.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  173:           blk.19.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  174:           blk.19.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  175:             blk.19.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  176:           blk.19.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  177:             blk.19.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  178:        blk.19.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  179:             blk.19.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  180:             blk.19.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  181:          blk.20.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  182:           blk.20.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  183:           blk.20.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  184:             blk.20.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  185:           blk.20.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  186:             blk.20.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  187:        blk.20.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  188:             blk.20.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  189:             blk.20.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  190:          blk.21.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  191:           blk.21.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  192:           blk.21.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  193:             blk.21.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  194:           blk.21.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  195:             blk.21.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  196:        blk.21.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  197:             blk.21.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  198:             blk.21.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  199:             blk.22.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  200:        blk.22.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  201:             blk.22.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  202:             blk.22.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  203:                    output.weight q6_K     [  4096, 32000,     1,     1 ]
llama_model_loader: - tensor  204:          blk.22.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  205:           blk.22.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  206:           blk.22.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  207:             blk.22.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  208:           blk.22.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  209:          blk.23.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  210:           blk.23.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  211:           blk.23.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  212:             blk.23.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  213:           blk.23.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  214:             blk.23.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  215:        blk.23.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  216:             blk.23.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  217:             blk.23.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  218:          blk.24.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  219:           blk.24.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  220:           blk.24.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  221:             blk.24.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  222:           blk.24.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  223:             blk.24.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  224:        blk.24.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  225:             blk.24.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  226:             blk.24.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  227:          blk.25.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  228:           blk.25.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  229:           blk.25.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  230:             blk.25.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  231:           blk.25.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  232:             blk.25.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  233:        blk.25.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  234:             blk.25.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  235:             blk.25.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  236:          blk.26.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  237:           blk.26.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  238:           blk.26.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  239:             blk.26.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  240:           blk.26.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  241:             blk.26.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  242:        blk.26.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  243:             blk.26.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  244:             blk.26.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  245:          blk.27.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  246:           blk.27.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  247:           blk.27.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  248:             blk.27.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  249:           blk.27.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  250:             blk.27.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  251:        blk.27.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  252:             blk.27.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  253:             blk.27.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  254:          blk.28.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  255:           blk.28.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  256:           blk.28.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  257:             blk.28.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  258:           blk.28.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  259:             blk.28.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  260:        blk.28.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  261:             blk.28.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  262:             blk.28.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  263:          blk.29.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  264:           blk.29.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  265:           blk.29.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  266:             blk.29.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  267:           blk.29.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  268:             blk.29.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  269:        blk.29.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  270:             blk.29.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  271:             blk.29.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  272:          blk.30.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  273:           blk.30.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  274:           blk.30.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  275:             blk.30.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  276:           blk.30.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  277:             blk.30.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  278:        blk.30.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  279:             blk.30.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  280:             blk.30.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  281:          blk.31.attn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  282:           blk.31.ffn_down.weight q4_0     [ 14336,  4096,     1,     1 ]
llama_model_loader: - tensor  283:           blk.31.ffn_gate.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  284:             blk.31.ffn_up.weight q4_0     [  4096, 14336,     1,     1 ]
llama_model_loader: - tensor  285:           blk.31.ffn_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: - tensor  286:             blk.31.attn_k.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  287:        blk.31.attn_output.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  288:             blk.31.attn_q.weight q4_0     [  4096,  4096,     1,     1 ]
llama_model_loader: - tensor  289:             blk.31.attn_v.weight q4_0     [  4096,  1024,     1,     1 ]
llama_model_loader: - tensor  290:               output_norm.weight f32      [  4096,     1,     1,     1 ]
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = llama
llama_model_loader: - kv   1:                               general.name str              = mistralai
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  11:                          general.file_type u32              = 2
llama_model_loader: - kv  12:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  13:                      tokenizer.ggml.tokens arr[str,32000]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  14:                      tokenizer.ggml.scores arr[f32,32000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,32000]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  16:                      tokenizer.ggml.merges arr[str,58980]   = ["▁ t", "i n", "e r", "▁ a", "h e...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  19:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  20:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  21:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  22:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  23:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type q4_0:  225 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens definition check successful ( 259/32000 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: n_ff             = 14336
llm_load_print_meta: n_expert         = 0
llm_load_print_meta: n_expert_used    = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 32768
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: model type       = 7B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 7.24 B
llm_load_print_meta: model size       = 3.83 GiB (4.54 BPW)
llm_load_print_meta: general.name     = mistralai
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
llm_load_tensors: ggml ctx size =    0.11 MiB
llm_load_tensors: mem required  = 3917.98 MiB
...................................................................................................
llama_new_context_with_model: n_ctx      = 2048
llama_new_context_with_model: freq_base  = 1000000.0
llama_new_context_with_model: freq_scale = 1
llama_new_context_with_model: KV self size  =  256.00 MiB, K (f16):  128.00 MiB, V (f16):  128.00 MiB
llama_build_graph: non-view tensors processed: 676/676
llama_new_context_with_model: compute buffer total size = 159.19 MiB
2024/01/19 21:40:12 ext_server_common.go:144: Starting internal llama main loop
[GIN] 2024/01/19 - 21:40:12 | 200 |  2.312708515s |       127.0.0.1 | POST     "/api/generate"
2024/01/19 21:40:55 ext_server_common.go:158: loaded 0 images
[GIN] 2024/01/19 - 21:41:45 | 200 | 50.293156053s |       127.0.0.1 | POST     "/api/generate"
<!-- gh-comment-id:1901657931 --> @Q-point commented on GitHub (Jan 20, 2024): I am following the same instructions here and I don't see it using the GPU. ``` dhq@dhq-desktop:~$ LD_LIBRARY_PATH=/usr/local/cuda/lib64 ollama serve 2024/01/19 21:25:40 images.go:808: total blobs: 8 2024/01/19 21:25:40 images.go:815: total unused blobs removed: 0 2024/01/19 21:25:40 routes.go:930: Listening on 127.0.0.1:11434 (version 0.1.20) 2024/01/19 21:25:41 shim_ext_server.go:142: Dynamic LLM variants [cuda] 2024/01/19 21:25:41 gpu.go:88: Detecting GPU type 2024/01/19 21:25:41 gpu.go:203: Searching for GPU management library libnvidia-ml.so 2024/01/19 21:25:41 gpu.go:248: Discovered GPU libraries: [] 2024/01/19 21:25:41 gpu.go:203: Searching for GPU management library librocm_smi64.so 2024/01/19 21:25:41 gpu.go:248: Discovered GPU libraries: [] 2024/01/19 21:25:41 routes.go:953: no GPU detected [GIN] 2024/01/19 - 21:26:07 | 200 | 151.903µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:26:07 | 404 | 493.628µs | 127.0.0.1 | POST "/api/show" [GIN] 2024/01/19 - 21:26:07 | 200 | 345.212605ms | 127.0.0.1 | POST "/api/pull" [GIN] 2024/01/19 - 21:26:26 | 200 | 76.704µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:26:26 | 404 | 233.598µs | 127.0.0.1 | POST "/api/show" [GIN] 2024/01/19 - 21:26:26 | 200 | 216.31751ms | 127.0.0.1 | POST "/api/pull" [GIN] 2024/01/19 - 21:26:42 | 200 | 72.608µs | 127.0.0.1 | HEAD "/" 2024/01/19 21:26:43 images.go:430: [model] - mistral 2024/01/19 21:26:44 download.go:123: downloading e8a35b5937a5 in 42 100 MB part(s) [GIN] 2024/01/19 - 21:31:18 | 200 | 88.32µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:31:18 | 200 | 14.408573ms | 127.0.0.1 | GET "/api/tags" [GIN] 2024/01/19 - 21:31:42 | 200 | 102.623µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:31:42 | 404 | 444.319µs | 127.0.0.1 | DELETE "/api/delete" [GIN] 2024/01/19 - 21:31:45 | 200 | 78.719µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:31:45 | 200 | 1.528636ms | 127.0.0.1 | DELETE "/api/delete" [GIN] 2024/01/19 - 21:31:50 | 200 | 87.455µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:31:50 | 200 | 880.67µs | 127.0.0.1 | GET "/api/tags" 2024/01/19 21:38:28 download.go:162: e8a35b5937a5 part 10 attempt 0 failed: unexpected EOF, retrying in 1s 2024/01/19 21:39:20 download.go:123: downloading 43070e2d4e53 in 1 11 KB part(s) 2024/01/19 21:39:23 download.go:123: downloading e6836092461f in 1 42 B part(s) 2024/01/19 21:39:26 download.go:123: downloading ed11eda7790d in 1 30 B part(s) 2024/01/19 21:39:28 download.go:123: downloading f9b1e3196ecf in 1 483 B part(s) 2024/01/19 21:39:36 images.go:430: [num_gpu] - 100 [GIN] 2024/01/19 - 21:39:36 | 200 | 12m53s | 127.0.0.1 | POST "/api/create" [GIN] 2024/01/19 - 21:40:10 | 200 | 85.567µs | 127.0.0.1 | HEAD "/" [GIN] 2024/01/19 - 21:40:10 | 200 | 2.897401ms | 127.0.0.1 | POST "/api/show" [GIN] 2024/01/19 - 21:40:10 | 200 | 1.145758ms | 127.0.0.1 | POST "/api/show" 2024/01/19 21:40:10 llm.go:71: GPU not available, falling back to CPU 2024/01/19 21:40:10 ext_server_common.go:136: Initializing internal llama server llama_model_loader: loaded meta data with 24 key-value pairs and 291 tensors from /home/dhq/.ollama/models/blobs/sha256:e8a35b5937a5e6d5c35d1f2a15f161e07eefe5e5bb0a3cdd42998ee79b057730 (version GGUF V3 (latest)) llama_model_loader: - tensor 0: token_embd.weight q4_0 [ 4096, 32000, 1, 1 ] llama_model_loader: - tensor 1: blk.0.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 2: blk.0.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 3: blk.0.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 4: blk.0.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 5: blk.0.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 6: blk.0.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 7: blk.0.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 8: blk.0.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 9: blk.0.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 10: blk.1.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 11: blk.1.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 12: blk.1.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 13: blk.1.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 14: blk.1.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 15: blk.1.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 16: blk.1.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 17: blk.1.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 18: blk.1.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 19: blk.10.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 20: blk.10.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 21: blk.10.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 22: blk.10.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 23: blk.10.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 24: blk.10.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 25: blk.2.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 26: blk.2.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 27: blk.2.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 28: blk.2.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 29: blk.2.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 30: blk.2.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 31: blk.2.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 32: blk.2.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 33: blk.2.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 34: blk.3.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 35: blk.3.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 36: blk.3.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 37: blk.3.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 38: blk.3.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 39: blk.3.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 40: blk.3.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 41: blk.3.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 42: blk.3.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 43: blk.4.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 44: blk.4.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 45: blk.4.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 46: blk.4.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 47: blk.4.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 48: blk.4.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 49: blk.4.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 50: blk.4.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 51: blk.4.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 52: blk.5.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 53: blk.5.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 54: blk.5.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 55: blk.5.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 56: blk.5.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 57: blk.5.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 58: blk.5.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 59: blk.5.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 60: blk.5.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 61: blk.6.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 62: blk.6.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 63: blk.6.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 64: blk.6.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 65: blk.6.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 66: blk.6.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 67: blk.6.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 68: blk.6.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 69: blk.6.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 70: blk.7.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 71: blk.7.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 72: blk.7.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 73: blk.7.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 74: blk.7.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 75: blk.7.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 76: blk.7.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 77: blk.7.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 78: blk.7.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 79: blk.8.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 80: blk.8.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 81: blk.8.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 82: blk.8.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 83: blk.8.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 84: blk.8.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 85: blk.8.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 86: blk.8.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 87: blk.8.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 88: blk.9.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 89: blk.9.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 90: blk.9.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 91: blk.9.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 92: blk.9.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 93: blk.9.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 94: blk.9.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 95: blk.9.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 96: blk.9.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 97: blk.10.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 98: blk.10.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 99: blk.10.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 100: blk.11.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 101: blk.11.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 102: blk.11.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 103: blk.11.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 104: blk.11.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 105: blk.11.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 106: blk.11.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 107: blk.11.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 108: blk.11.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 109: blk.12.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 110: blk.12.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 111: blk.12.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 112: blk.12.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 113: blk.12.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 114: blk.12.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 115: blk.12.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 116: blk.12.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 117: blk.12.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 118: blk.13.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 119: blk.13.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 120: blk.13.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 121: blk.13.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 122: blk.13.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 123: blk.13.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 124: blk.13.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 125: blk.13.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 126: blk.13.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 127: blk.14.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 128: blk.14.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 129: blk.14.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 130: blk.14.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 131: blk.14.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 132: blk.14.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 133: blk.14.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 134: blk.14.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 135: blk.14.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 136: blk.15.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 137: blk.15.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 138: blk.15.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 139: blk.15.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 140: blk.15.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 141: blk.15.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 142: blk.15.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 143: blk.15.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 144: blk.15.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 145: blk.16.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 146: blk.16.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 147: blk.16.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 148: blk.16.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 149: blk.16.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 150: blk.16.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 151: blk.16.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 152: blk.16.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 153: blk.16.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 154: blk.17.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 155: blk.17.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 156: blk.17.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 157: blk.17.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 158: blk.17.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 159: blk.17.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 160: blk.17.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 161: blk.17.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 162: blk.17.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 163: blk.18.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 164: blk.18.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 165: blk.18.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 166: blk.18.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 167: blk.18.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 168: blk.18.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 169: blk.18.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 170: blk.18.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 171: blk.18.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 172: blk.19.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 173: blk.19.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 174: blk.19.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 175: blk.19.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 176: blk.19.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 177: blk.19.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 178: blk.19.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 179: blk.19.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 180: blk.19.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 181: blk.20.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 182: blk.20.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 183: blk.20.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 184: blk.20.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 185: blk.20.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 186: blk.20.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 187: blk.20.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 188: blk.20.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 189: blk.20.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 190: blk.21.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 191: blk.21.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 192: blk.21.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 193: blk.21.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 194: blk.21.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 195: blk.21.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 196: blk.21.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 197: blk.21.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 198: blk.21.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 199: blk.22.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 200: blk.22.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 201: blk.22.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 202: blk.22.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 203: output.weight q6_K [ 4096, 32000, 1, 1 ] llama_model_loader: - tensor 204: blk.22.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 205: blk.22.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 206: blk.22.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 207: blk.22.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 208: blk.22.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 209: blk.23.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 210: blk.23.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 211: blk.23.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 212: blk.23.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 213: blk.23.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 214: blk.23.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 215: blk.23.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 216: blk.23.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 217: blk.23.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 218: blk.24.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 219: blk.24.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 220: blk.24.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 221: blk.24.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 222: blk.24.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 223: blk.24.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 224: blk.24.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 225: blk.24.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 226: blk.24.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 227: blk.25.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 228: blk.25.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 229: blk.25.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 230: blk.25.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 231: blk.25.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 232: blk.25.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 233: blk.25.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 234: blk.25.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 235: blk.25.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 236: blk.26.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 237: blk.26.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 238: blk.26.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 239: blk.26.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 240: blk.26.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 241: blk.26.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 242: blk.26.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 243: blk.26.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 244: blk.26.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 245: blk.27.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 246: blk.27.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 247: blk.27.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 248: blk.27.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 249: blk.27.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 250: blk.27.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 251: blk.27.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 252: blk.27.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 253: blk.27.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 254: blk.28.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 255: blk.28.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 256: blk.28.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 257: blk.28.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 258: blk.28.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 259: blk.28.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 260: blk.28.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 261: blk.28.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 262: blk.28.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 263: blk.29.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 264: blk.29.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 265: blk.29.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 266: blk.29.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 267: blk.29.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 268: blk.29.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 269: blk.29.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 270: blk.29.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 271: blk.29.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 272: blk.30.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 273: blk.30.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 274: blk.30.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 275: blk.30.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 276: blk.30.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 277: blk.30.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 278: blk.30.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 279: blk.30.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 280: blk.30.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 281: blk.31.attn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 282: blk.31.ffn_down.weight q4_0 [ 14336, 4096, 1, 1 ] llama_model_loader: - tensor 283: blk.31.ffn_gate.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 284: blk.31.ffn_up.weight q4_0 [ 4096, 14336, 1, 1 ] llama_model_loader: - tensor 285: blk.31.ffn_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: - tensor 286: blk.31.attn_k.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 287: blk.31.attn_output.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 288: blk.31.attn_q.weight q4_0 [ 4096, 4096, 1, 1 ] llama_model_loader: - tensor 289: blk.31.attn_v.weight q4_0 [ 4096, 1024, 1, 1 ] llama_model_loader: - tensor 290: output_norm.weight f32 [ 4096, 1, 1, 1 ] llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = llama llama_model_loader: - kv 1: general.name str = mistralai llama_model_loader: - kv 2: llama.context_length u32 = 32768 llama_model_loader: - kv 3: llama.embedding_length u32 = 4096 llama_model_loader: - kv 4: llama.block_count u32 = 32 llama_model_loader: - kv 5: llama.feed_forward_length u32 = 14336 llama_model_loader: - kv 6: llama.rope.dimension_count u32 = 128 llama_model_loader: - kv 7: llama.attention.head_count u32 = 32 llama_model_loader: - kv 8: llama.attention.head_count_kv u32 = 8 llama_model_loader: - kv 9: llama.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 10: llama.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 11: general.file_type u32 = 2 llama_model_loader: - kv 12: tokenizer.ggml.model str = llama llama_model_loader: - kv 13: tokenizer.ggml.tokens arr[str,32000] = ["<unk>", "<s>", "</s>", "<0x00>", "<... llama_model_loader: - kv 14: tokenizer.ggml.scores arr[f32,32000] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 15: tokenizer.ggml.token_type arr[i32,32000] = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ... llama_model_loader: - kv 16: tokenizer.ggml.merges arr[str,58980] = ["▁ t", "i n", "e r", "▁ a", "h e... llama_model_loader: - kv 17: tokenizer.ggml.bos_token_id u32 = 1 llama_model_loader: - kv 18: tokenizer.ggml.eos_token_id u32 = 2 llama_model_loader: - kv 19: tokenizer.ggml.unknown_token_id u32 = 0 llama_model_loader: - kv 20: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 21: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 22: tokenizer.chat_template str = {{ bos_token }}{% for message in mess... llama_model_loader: - kv 23: general.quantization_version u32 = 2 llama_model_loader: - type f32: 65 tensors llama_model_loader: - type q4_0: 225 tensors llama_model_loader: - type q6_K: 1 tensors llm_load_vocab: special tokens definition check successful ( 259/32000 ). llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = llama llm_load_print_meta: vocab type = SPM llm_load_print_meta: n_vocab = 32000 llm_load_print_meta: n_merges = 0 llm_load_print_meta: n_ctx_train = 32768 llm_load_print_meta: n_embd = 4096 llm_load_print_meta: n_head = 32 llm_load_print_meta: n_head_kv = 8 llm_load_print_meta: n_layer = 32 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_gqa = 4 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-05 llm_load_print_meta: f_clamp_kqv = 0.0e+00 llm_load_print_meta: f_max_alibi_bias = 0.0e+00 llm_load_print_meta: n_ff = 14336 llm_load_print_meta: n_expert = 0 llm_load_print_meta: n_expert_used = 0 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 1000000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_yarn_orig_ctx = 32768 llm_load_print_meta: rope_finetuned = unknown llm_load_print_meta: model type = 7B llm_load_print_meta: model ftype = Q4_0 llm_load_print_meta: model params = 7.24 B llm_load_print_meta: model size = 3.83 GiB (4.54 BPW) llm_load_print_meta: general.name = mistralai llm_load_print_meta: BOS token = 1 '<s>' llm_load_print_meta: EOS token = 2 '</s>' llm_load_print_meta: UNK token = 0 '<unk>' llm_load_print_meta: LF token = 13 '<0x0A>' llm_load_tensors: ggml ctx size = 0.11 MiB llm_load_tensors: mem required = 3917.98 MiB ................................................................................................... llama_new_context_with_model: n_ctx = 2048 llama_new_context_with_model: freq_base = 1000000.0 llama_new_context_with_model: freq_scale = 1 llama_new_context_with_model: KV self size = 256.00 MiB, K (f16): 128.00 MiB, V (f16): 128.00 MiB llama_build_graph: non-view tensors processed: 676/676 llama_new_context_with_model: compute buffer total size = 159.19 MiB 2024/01/19 21:40:12 ext_server_common.go:144: Starting internal llama main loop [GIN] 2024/01/19 - 21:40:12 | 200 | 2.312708515s | 127.0.0.1 | POST "/api/generate" 2024/01/19 21:40:55 ext_server_common.go:158: loaded 0 images [GIN] 2024/01/19 - 21:41:45 | 200 | 50.293156053s | 127.0.0.1 | POST "/api/generate" ```
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#664
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?