[Finetune] Integrate Chat template #178
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
carsonwang
requested changes
Apr 9, 2024
There was a problem hiding this comment.
Thanks for the work! Summarized the changes to update as we discussed offline:
- Remove the added
is_base_modelparameter in the finetuning yaml file. - Allow user configuring
chat_templatein the yaml file. In most of the case, people don't configure it. Priority order: user configured chat_template > model's chat_template > our default template - Write the default template by following other models' (such as llama2 chat), that is, check roles in the message, etc.
- The original data format needs to convert to chat format first, before applying the chat template.
- Add unit tests to test the result after applying chat template, covering all use cases.
- Support chat format as finetuning dataset format. Please follow openAI's format. We can support this in a separate PR.
minmingzhu
force-pushed
the
chat_template
branch
from
3e6ccac to
6a0bf63
Compare
harborn
reviewed
Apr 18, 2024
| @@ -15,6 +15,7 @@ The following are the parameters supported in the finetuning workflow. | |||
| |lora_config|task_type: CAUSAL_LM<br>r: 8<br>lora_alpha: 32<br>lora_dropout: 0.1|Will be passed to the LoraConfig `__init__()` method, then it'll be used as config to build Peft model object.| | |||
| |deltatuner_config|"algo": "lora"<br>"denas": True<br>"best_model_structure": "/path/to/best_structure_of_deltatuner_model"|Will be passed to the DeltaTunerArguments `__init__()` method, then it'll be used as config to build [Deltatuner model](https://github.com/intel/e2eAIOK/tree/main/e2eAIOK/deltatuner) object.| | |||
| |enable_gradient_checkpointing|False|enable gradient checkpointing to save GPU memory, but will cost more compute runtime| | |||
| |chat_template|None|User-defined chat template.| | |||
There was a problem hiding this comment.
Have you compared the impact of different templates on fine-tuning performance?
minmingzhu
force-pushed
the
chat_template
branch
from
42825d3 to
4fa89cc
Compare
Signed-off-by: minmingzhu <minming.zhu@intel.com>
Signed-off-by: minmingzhu <minming.zhu@intel.com>
2. modify chat template Signed-off-by: minmingzhu <minming.zhu@intel.com>
2. add unit test Signed-off-by: minmingzhu <minming.zhu@intel.com>
* update * fix blocking * update Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * update Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * fix setup and getting started Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * update Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * update Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * nit Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * Add dependencies for tests and update pyproject.toml Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * Update dependencies and test workflow Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * Update dependencies and fix torch_dist.py Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> * Update OpenAI SDK installation and start ray cluster Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com> --------- Signed-off-by: Wu, Xiaochang <xiaochang.wu@intel.com>
* single test * single test * single test * single test * fix hang error
* use base model mpt-7b instead of mpt-7b-chat Signed-off-by: minmingzhu <minming.zhu@intel.com> * manual setting specify tokenizer Signed-off-by: minmingzhu <minming.zhu@intel.com> * update Signed-off-by: minmingzhu <minming.zhu@intel.com> * update doc/finetune_parameters.md Signed-off-by: minmingzhu <minming.zhu@intel.com> --------- Signed-off-by: minmingzhu <minming.zhu@intel.com>
Signed-off-by: minmingzhu <minming.zhu@intel.com>
Signed-off-by: minmingzhu <minming.zhu@intel.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
No description provided.