github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-07 00:22:43 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #1496] Add Phi-2 model #47320

New Issue
Closed
opened 2026-04-28 03:35:13 -05:00 by GiteaMirror · 19 comments
GiteaMirror commented 2026-04-28 03:35:13 -05:00
Owner
Copy Link

Originally created by @hunnble on GitHub (Dec 13, 2023).
Original GitHub issue: https://github.com/ollama/ollama/issues/1496

Originally assigned to: @jmorganca on GitHub.

The Phi-2 model performs well. Should we consider adding it to Ollama?

Originally created by @hunnble on GitHub (Dec 13, 2023). Original GitHub issue: https://github.com/ollama/ollama/issues/1496 Originally assigned to: @jmorganca on GitHub. The Phi-2 model performs well. Should we consider adding it to Ollama?
GiteaMirror commented 2026-04-28 03:35:14 -05:00
Author
Owner
Copy Link

@easp commented on GitHub (Dec 13, 2023):

It appears to use the same model architecture as Phi-1.5. That model architecure is not supported by Llama.cpp.

In addition, the model weights are not licensed for redistribution, at least not currently. You have to log-in to Azure with an account that has a credit card on file.

<!-- gh-comment-id:1854408073 --> @easp commented on GitHub (Dec 13, 2023): It appears to use the same model architecture as Phi-1.5. That [model architecure is not supported by Llama.cpp](https://github.com/ggerganov/llama.cpp/issues/3146). In addition, the model weights are not licensed for redistribution, at least not currently. You have to log-in to Azure with an account that has a credit card on file.
GiteaMirror commented 2026-04-28 03:35:14 -05:00
Author
Owner
Copy Link

@izard commented on GitHub (Dec 14, 2023):

The model weights can be downloaded by anyone from official https://huggingface.co/microsoft/phi-2/tree/main MSFT page. The license is quite permissive.

<!-- gh-comment-id:1854939717 --> @izard commented on GitHub (Dec 14, 2023): The model weights can be downloaded by anyone from official https://huggingface.co/microsoft/phi-2/tree/main MSFT page. The license is quite permissive.
GiteaMirror commented 2026-04-28 03:35:14 -05:00
Author
Owner
Copy Link

@easp commented on GitHub (Dec 14, 2023):

They've loosened things up in the last 10hr. The fact remains though that support for the model architecture in Llama.cpp hasn't progressed much (at least not visibly) since September.

<!-- gh-comment-id:1855084968 --> @easp commented on GitHub (Dec 14, 2023): They've loosened things up in the last 10hr. The fact remains though that support for the model architecture in Llama.cpp hasn't progressed much (at least not visibly) since September.
GiteaMirror commented 2026-04-28 03:35:15 -05:00
Author
Owner
Copy Link

@HardKothari commented on GitHub (Dec 15, 2023):

Not sure if this can help but there is a model card with GGUF format file and if they can be used. I tried locally creating Modelfile and it had internal error in the server.

https://huggingface.co/radames/phi-2-quantized

<!-- gh-comment-id:1858487772 --> @HardKothari commented on GitHub (Dec 15, 2023): Not sure if this can help but there is a model card with GGUF format file and if they can be used. I tried locally creating Modelfile and it had internal error in the server. https://huggingface.co/radames/phi-2-quantized
GiteaMirror commented 2026-04-28 03:35:15 -05:00
Author
Owner
Copy Link

@jingfelix commented on GitHub (Dec 16, 2023):

Seems like Phi-2 will be supported by llama.cpp in forseeeable future 👀
https://github.com/ggerganov/llama.cpp/pull/4490

<!-- gh-comment-id:1858712414 --> @jingfelix commented on GitHub (Dec 16, 2023): Seems like Phi-2 will be supported by llama.cpp in forseeeable future 👀 https://github.com/ggerganov/llama.cpp/pull/4490
GiteaMirror commented 2026-04-28 03:35:15 -05:00
Author
Owner
Copy Link

@vaclcer commented on GitHub (Dec 18, 2023):

it is now. Fingers crossed for Ollama :)

<!-- gh-comment-id:1861732403 --> @vaclcer commented on GitHub (Dec 18, 2023): it is now. Fingers crossed for Ollama :)
GiteaMirror commented 2026-04-28 03:35:16 -05:00
Author
Owner
Copy Link

@Jbollenbacher commented on GitHub (Dec 19, 2023):

Seems like Phi-2 will be supported by llama.cpp in forseeeable future 👀
https://github.com/ggerganov/llama.cpp/pull/4490

This just merged. llama.cpp now supports phi-2.

<!-- gh-comment-id:1862009956 --> @Jbollenbacher commented on GitHub (Dec 19, 2023): > Seems like Phi-2 will be supported by llama.cpp in forseeeable future 👀 > https://github.com/ggerganov/llama.cpp/pull/4490 This just merged. llama.cpp now supports phi-2.
GiteaMirror commented 2026-04-28 03:35:16 -05:00
Author
Owner
Copy Link

@brianjking commented on GitHub (Dec 19, 2023):

Now that this has been merged what is required to get this working in ollama?

<!-- gh-comment-id:1862973613 --> @brianjking commented on GitHub (Dec 19, 2023): Now that this has been merged what is required to get this working in ollama?
GiteaMirror commented 2026-04-28 03:35:17 -05:00
Author
Owner
Copy Link

@HardKothari commented on GitHub (Dec 19, 2023):

I believe Ollama has already added the support and is already available in pre-release v0.1.17
Check this out..

https://github.com/jmorganca/ollama/releases/tag/v0.1.17

https://ollama.ai/jmorgan/phi/tags

<!-- gh-comment-id:1863037480 --> @HardKothari commented on GitHub (Dec 19, 2023): I believe Ollama has already added the support and is already available in pre-release v0.1.17 Check this out.. https://github.com/jmorganca/ollama/releases/tag/v0.1.17 https://ollama.ai/jmorgan/phi/tags
GiteaMirror commented 2026-04-28 03:35:17 -05:00
Author
Owner
Copy Link

@ibnbd commented on GitHub (Dec 20, 2023):

https://ollama.ai/library/phi

<!-- gh-comment-id:1864501642 --> @ibnbd commented on GitHub (Dec 20, 2023): https://ollama.ai/library/phi
GiteaMirror commented 2026-04-28 03:35:17 -05:00
Author
Owner
Copy Link

@oliverbob commented on GitHub (Dec 23, 2023):

Yes, its lightning fast. Its already supported by ollama.

<!-- gh-comment-id:1868267389 --> @oliverbob commented on GitHub (Dec 23, 2023): Yes, its lightning fast. Its already supported by ollama.
GiteaMirror commented 2026-04-28 03:35:17 -05:00
Author
Owner
Copy Link

@snajjar commented on GitHub (Dec 23, 2023):

Not working on my machine, :(
(Lenovo Thinkpad P14S, Linux 6.5.7 Arch)

$ ollama run phi
⠙   Error: llama runner: failed to load model '/usr/share/ollama/.ollama/models/blobs/sha256:bd608f9545597ea3278b78038943059d1c29c62f3ca02c86523014f3a8c7a7f1': this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull phi:latest

I tried different tags for the phi model, but none of them work. Do I need to install ollama from git?

<!-- gh-comment-id:1868278397 --> @snajjar commented on GitHub (Dec 23, 2023): Not working on my machine, :( (Lenovo Thinkpad P14S, Linux 6.5.7 Arch) ``` $ ollama run phi ⠙ Error: llama runner: failed to load model '/usr/share/ollama/.ollama/models/blobs/sha256:bd608f9545597ea3278b78038943059d1c29c62f3ca02c86523014f3a8c7a7f1': this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull phi:latest ``` I tried different tags for the phi model, but none of them work. Do I need to install ollama from git?
GiteaMirror commented 2026-04-28 03:35:18 -05:00
Author
Owner
Copy Link

@bgokden commented on GitHub (Dec 23, 2023):

Not working on my machine, :( (Lenovo Thinkpad P14S, Linux 6.5.7 Arch)

$ ollama run phi
⠙   Error: llama runner: failed to load model '/usr/share/ollama/.ollama/models/blobs/sha256:bd608f9545597ea3278b78038943059d1c29c62f3ca02c86523014f3a8c7a7f1': this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull phi:latest

I tried different tags for the phi model, but none of them work. Do I need to install ollama from git?

Which version of ollama are you using? It worked for me with v0.1.17.
I have installed ollama from git with cloning this tag.

<!-- gh-comment-id:1868279600 --> @bgokden commented on GitHub (Dec 23, 2023): > Not working on my machine, :( (Lenovo Thinkpad P14S, Linux 6.5.7 Arch) > > ``` > $ ollama run phi > ⠙ Error: llama runner: failed to load model '/usr/share/ollama/.ollama/models/blobs/sha256:bd608f9545597ea3278b78038943059d1c29c62f3ca02c86523014f3a8c7a7f1': this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull phi:latest > ``` > > I tried different tags for the phi model, but none of them work. Do I need to install ollama from git? Which version of ollama are you using? It worked for me with v0.1.17. I have installed ollama from git with cloning this tag.
GiteaMirror commented 2026-04-28 03:35:18 -05:00
Author
Owner
Copy Link

@snajjar commented on GitHub (Dec 23, 2023):

Indeed. Even though the package version stated 1.0.17, ollama --version gave 1.0.9...

Re-installed from git, it works in 1.0.17

<!-- gh-comment-id:1868294768 --> @snajjar commented on GitHub (Dec 23, 2023): Indeed. Even though the package version stated 1.0.17, `ollama --version` gave 1.0.9... Re-installed from git, it works in 1.0.17
GiteaMirror commented 2026-04-28 03:35:18 -05:00
Author
Owner
Copy Link

@oliverbob commented on GitHub (Dec 23, 2023):

Not working on my machine, :( (Lenovo Thinkpad P14S, Linux 6.5.7 Arch)

$ ollama run phi
⠙   Error: llama runner: failed to load model '/usr/share/ollama/.ollama/models/blobs/sha256:bd608f9545597ea3278b78038943059d1c29c62f3ca02c86523014f3a8c7a7f1': this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull phi:latest

I tried different tags for the phi model, but none of them work. Do I need to install ollama from git?

You clearly need an update if that's the case. Just run the installer via curl https://ollama.ai/install.sh | sh to update it once more.

<!-- gh-comment-id:1868317913 --> @oliverbob commented on GitHub (Dec 23, 2023): > Not working on my machine, :( (Lenovo Thinkpad P14S, Linux 6.5.7 Arch) > > ``` > $ ollama run phi > ⠙ Error: llama runner: failed to load model '/usr/share/ollama/.ollama/models/blobs/sha256:bd608f9545597ea3278b78038943059d1c29c62f3ca02c86523014f3a8c7a7f1': this model may be incompatible with your version of Ollama. If you previously pulled this model, try updating it by running `ollama pull phi:latest > ``` > > I tried different tags for the phi model, but none of them work. Do I need to install ollama from git? You clearly need an update if that's the case. Just run the installer via `curl https://ollama.ai/install.sh | sh` to update it once more.
GiteaMirror commented 2026-04-28 03:35:19 -05:00
Author
Owner
Copy Link

@oliverbob commented on GitHub (Dec 23, 2023):

Indeed. Even though the package version stated 1.0.17, ollama --version gave 1.0.9...

Re-installed from git, it works in 1.0.17

You could be running two versions of Ollama. One via docker and one via the install script.

Run sudo lsof -i :11434 to find out. Or docker ps. If its on docker, you need to rebuild docker, but only when you have don a fresh git pull to update the local repo. Either remove docker ollama or stop it in docker before installing the curl https://ollama.ai/install.sh | sh. Then redownload Phi-2.

<!-- gh-comment-id:1868318757 --> @oliverbob commented on GitHub (Dec 23, 2023): > Indeed. Even though the package version stated 1.0.17, `ollama --version` gave 1.0.9... > > Re-installed from git, it works in 1.0.17 You could be running two versions of Ollama. One via docker and one via the install script. Run sudo lsof -i :11434 to find out. Or `docker ps`. If its on docker, you need to rebuild docker, but only when you have don a fresh `git pull` to update the local repo. Either remove docker ollama or stop it in docker before installing the `curl https://ollama.ai/install.sh | sh`. Then redownload Phi-2.
GiteaMirror commented 2026-04-28 03:35:19 -05:00
Author
Owner
Copy Link

@trickster commented on GitHub (Dec 24, 2023):

Is there any plan to add dolphin phi-2 as well? https://huggingface.co/cognitivecomputations/dolphin-2_6-phi-2/tree/main

<!-- gh-comment-id:1868430989 --> @trickster commented on GitHub (Dec 24, 2023): Is there any plan to add dolphin phi-2 as well? https://huggingface.co/cognitivecomputations/dolphin-2_6-phi-2/tree/main
GiteaMirror commented 2026-04-28 03:35:19 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Dec 24, 2023):

@trickster yes! It's here https://ollama.ai/library/dolphin-phi 😊

<!-- gh-comment-id:1868601001 --> @jmorganca commented on GitHub (Dec 24, 2023): @trickster yes! It's here https://ollama.ai/library/dolphin-phi 😊
GiteaMirror commented 2026-04-28 03:35:20 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Dec 24, 2023):

Closing this as Phi has been added!

<!-- gh-comment-id:1868601066 --> @jmorganca commented on GitHub (Dec 24, 2023): Closing this as Phi has been added!
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#47320
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?