github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-05-06 08:02:14 -05:00
Code Issues 14.5k Packages Projects Releases 120 Wiki Activity

[GH-ISSUE #7211] How to get forward method of model #66635

New Issue
Closed
opened 2026-05-04 07:39:58 -05:00 by GiteaMirror · 7 comments
GiteaMirror commented 2026-05-04 07:39:58 -05:00
Owner
Copy Link

Originally created by @VijayRajIITP on GitHub (Oct 15, 2024).
Original GitHub issue: https://github.com/ollama/ollama/issues/7211

What is the issue?

I am using ollama
ollama pull llama3
llm=Ollama(model='llama3')
base_url="http://localhost:11434/v1"
l want to work with foraward method of model how to get that .Is it poosible

OS

Linux

GPU

Nvidia

CPU

No response

Ollama version

No response

Originally created by @VijayRajIITP on GitHub (Oct 15, 2024). Original GitHub issue: https://github.com/ollama/ollama/issues/7211 ### What is the issue? I am using ollama ollama pull llama3 llm=Ollama(model='llama3') base_url="http://localhost:11434/v1" l want to work with foraward method of model how to get that .Is it poosible ### OS Linux ### GPU Nvidia ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the question label 2026-05-04 07:39:58 -05:00
GiteaMirror commented 2026-05-04 07:40:11 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 15, 2024):

What is forward method?

<!-- gh-comment-id:2414694836 --> @rick-github commented on GitHub (Oct 15, 2024): What is `forward method`?
GiteaMirror commented 2026-05-04 07:40:11 -05:00
Author
Owner
Copy Link

@VijayRajIITP commented on GitHub (Oct 15, 2024):

function that computes the output of the model given an input. It encapsulates the entire forward pass through the layers of the model, processing the input data and producing predictions or embeddings.
Encoding and decoding

<!-- gh-comment-id:2414759621 --> @VijayRajIITP commented on GitHub (Oct 15, 2024): function that computes the output of the model given an input. It encapsulates the entire forward pass through the layers of the model, processing the input data and producing predictions or embeddings. Encoding and decoding
GiteaMirror commented 2026-05-04 07:40:12 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 15, 2024):

Are you referring to logprobs? https://github.com/ollama/ollama/issues/2415

<!-- gh-comment-id:2414852509 --> @rick-github commented on GitHub (Oct 15, 2024): Are you referring to logprobs? https://github.com/ollama/ollama/issues/2415
GiteaMirror commented 2026-05-04 07:40:13 -05:00
Author
Owner
Copy Link

@VijayRajIITP commented on GitHub (Oct 18, 2024):

from langchain.from langchain.llms import Ollama
llm=Ollama(model='llama3')
The response going through rest api using post method , can we acess server side code as it is local host , I want to see the internal working how my input is going to lma model and all the internal detail of model in .py itself

<!-- gh-comment-id:2421410315 --> @VijayRajIITP commented on GitHub (Oct 18, 2024): from langchain.from langchain.llms import Ollama llm=Ollama(model='llama3') The response going through rest api using post method , can we acess server side code as it is local host , I want to see the internal working how my input is going to lma model and all the internal detail of model in .py itself
GiteaMirror commented 2026-05-04 07:40:14 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 18, 2024):

You can use strace to see the internal detail:

strace python3 my_ollama_script.py
<!-- gh-comment-id:2422180579 --> @rick-github commented on GitHub (Oct 18, 2024): You can use `strace` to see the internal detail: ``` strace python3 my_ollama_script.py ```
GiteaMirror commented 2026-05-04 07:40:15 -05:00
Author
Owner
Copy Link

@VijayRajIITP commented on GitHub (Oct 21, 2024):

I’m trying to understand how Ollama manages tokenization and detokenization when interacting with APIs. I used debug tools to inspect the process, but the responses received from the API are still in text form. I’m interested in the internal workings that occur after a response is processed through the API, such as how tokenization and detokenization are handled.

I also explored the Ollama installation directory, which I set up using the command:

curl -fsSL https://ollama.com/install.sh | sh

However, I couldn’t find any files related to the model there. This raises the question of how the model is loaded, how it processes the input, and how the output is post-processed.

<!-- gh-comment-id:2426491452 --> @VijayRajIITP commented on GitHub (Oct 21, 2024): I’m trying to understand how Ollama manages tokenization and detokenization when interacting with APIs. I used debug tools to inspect the process, but the responses received from the API are still in text form. I’m interested in the internal workings that occur after a response is processed through the API, such as how tokenization and detokenization are handled. I also explored the Ollama installation directory, which I set up using the command: ```bash curl -fsSL https://ollama.com/install.sh | sh ``` However, I couldn’t find any files related to the model there. This raises the question of how the model is loaded, how it processes the input, and how the output is post-processed.
GiteaMirror commented 2026-05-04 07:40:16 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Oct 21, 2024):

ollama is a wrapper for llama.cpp, you can find the details about the internal workings there.

<!-- gh-comment-id:2426725095 --> @rick-github commented on GitHub (Oct 21, 2024): ollama is a wrapper for [llama.cpp](https://github.com/ggerganov/llama.cpp), you can find the details about the internal workings there.
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 question
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#66635
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?