Gemma3 initial commit #102
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Updated configs Moved embed_tokens to more natural place Updated run_model to use sampler and stop at end_of_turn token Added test_sharding_gemma3 Added batched forward test. Need more complex behavior and testing
jenriver
reviewed
Dec 17, 2025
jenriver
reviewed
Dec 17, 2025
jenriver
reviewed
Dec 17, 2025
Member
|
Also, please make sure your selective tests are passing |
jenriver
reviewed
Dec 22, 2025
Comment on lines
32
to
60
| def make_input(processor, dtype=torch.float32, msg1=True): | ||
| if msg1: | ||
| messages = [ | ||
| {"role": "system", "content": [{"type": "text", "text": "You are a helpful assistant."}]}, | ||
| { | ||
| "role": "user", | ||
| "content": [ | ||
| { | ||
| "type": "image", | ||
| "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg", | ||
| }, | ||
| {"type": "text", "text": "What is shown in this image?"}, | ||
| ], | ||
| }, | ||
| ] | ||
| else: | ||
| messages = [ | ||
| {"role": "system", "content": [{"type": "text", "text": "You are a helpful assistant."}]}, | ||
| { | ||
| "role": "user", | ||
| "content": [ | ||
| { | ||
| "type": "image", | ||
| "image": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg", | ||
| }, | ||
| {"type": "text", "text": "Describe this image in detail."}, | ||
| ], | ||
| }, | ||
| ] |
Member
There was a problem hiding this comment.
This is very lengthy and difficult to read. How about having something like this?
def make_input(processor, dtype=torch.float32, msg1=True):
url = "pipeline-cat-chonk.jpeg" if msg1 else "bee.jpg"
prompt = "What is shown in this image?" if msg1 else "Describe this image in detail."
img_key = "url" if msg1 else "image"
messages = [
{"role": "system", "content": [{"type": "text", "text": "You are a helpful assistant."}]},
{"role": "user", "content": [
{"type": "image", img_key: f"https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/{url}"},
{"type": "text", "text": prompt}
]}
]
# Add your return statement or processor logic here
Collaborator
Author
There was a problem hiding this comment.
I implemented your suggested change. The final output ends up longer because of the ruff formatting, but it is shorter than before.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Resolves #100
Adds the Gemma3 model to the Bonsai Repo. This first commit is just a working version, but I am still working on optimizing it.
Reference
Refer to Issue #100
Checklist
run_model.pyfor model usage,test_outputs.pyand/ormodel_validation_colab.ipynbfor quality).