Skip to content

Mistral 3 Fast Inference#289

Merged
danielhanchen merged 85 commits into
unslothai:mainfrom
Datta0:mistral3_vllm
Sep 22, 2025
Merged

Mistral 3 Fast Inference#289
danielhanchen merged 85 commits into
unslothai:mainfrom
Datta0:mistral3_vllm

Conversation

@Datta0
Copy link
Copy Markdown
Collaborator

@Datta0 Datta0 commented Sep 17, 2025

No description provided.

# Find missing attributes (in original but not in new)
missing_in_new = orig_attrs - new_attrs
missing_in_new = missing_in_new - {'hf_device_map'}
missing_in_new = missing_in_new - {'hf_device_map', 'source_cls'}
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

if this is a set, wrap in set ie set({...})

Copy link
Copy Markdown
Collaborator Author

@Datta0 Datta0 Sep 21, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh this is already a set right? I don't think we need to do a wrap


# Get layer names from config
layer_templates = get_model_layer_config(return_non_layered=False)
layer_names = sum(layer_templates.values(), [])
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wait sum?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh basically this is like join but I wanted a single deflated list. So sum everything with empty list and you get a single list of everything :)

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

its like a=[1,2,3]; b = [4,5,6] and doing a+b which is [1,2,3,4,5,6] or something

Copy link
Copy Markdown
Member

@danielhanchen danielhanchen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Nice work!

@Datta0 Datta0 marked this pull request as ready for review September 21, 2025 08:31
@Datta0 Datta0 changed the title [WIP] Mistral 3 Fast Inference Mistral 3 Fast Inference Sep 21, 2025
@danielhanchen danielhanchen merged commit 661df44 into unslothai:main Sep 22, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants