Open
Conversation
Add auto module with composable base for loading models from HuggingFace Hub. - AutoConfig: loads config.json, parses model_type - Weights: loads safetensors (single or sharded) - Model + CausalLM traits - AutoModelForCausalLM: factory returning Box<dyn CausalLM> Supports: llama, mistral, phi3, qwen2, gemma
Add Model and CausalLM trait implementations to: - llama (LlamaForCausalLM wrapper) - mistral - phi3 - qwen2 - gemma
…39/candle into feature/auto-model-causal-lm-3120
Contributor
Author
|
Looping in maintainer for review, thank you! |
Member
|
Don't worry I was already looped in - just haven't had the time yet :) |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Closes #3120
I took inspiration from Python HuggingFace Transformers. Since candle is also a HuggingFace project, I use the same naming (AutoModelForCausalLM, from_pretrained) so users familiar with the Python API feel at home.
The shared base (AutoConfig, Weights) is designed so subsequent PRs can add AutoModelForSequenceClassification, AutoModelForMaskedLM, etc. if maintainers agree with this direction.
Currently supports
llama, mistral, phi3, qwen2, gemma
Suggested review path
Commit 1 - auto/ module f35d137
Commit 2 - trait impls in model files 6336ff6
Commit 3 - example a9c086f