Revised PoSE #8822
Merged
Revised PoSE #8822
Conversation
This file contains 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
|
Thanks for your contribution! |
|
|
lugimzzz
reviewed
Sep 24, 2024
llm/run_finetune.py
Outdated
| model_config.long_sequence_init_args = { | ||
| "dim": int(model_config.hidden_size / model_config.num_attention_heads), | ||
| "max_position_embeddings": data_args.scaled_max_length, # extended context window | ||
| "base": 10000, |
llm/run_finetune.py
Outdated
| if ptq_ds is not None | ||
| else None | ||
| ) | ||
| train_ds = train_ds.map(partial(trans_func)) if train_ds is not None else None |
llm/utils/data.py
Outdated
| return features | ||
|
|
||
|
|
||
| def test_preprocess_function(example, tokenizer, inference_length): |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## develop #8822 +/- ##
===========================================
+ Coverage 52.81% 52.91% +0.10%
===========================================
Files 677 679 +2
Lines 107943 108403 +460
===========================================
+ Hits 57010 57365 +355
- Misses 50933 51038 +105 ☔ View full report in Codecov by Sentry. |
lugimzzz
reviewed
Oct 21, 2024
llm/run_finetune.py
Outdated
| from utils.data import convert_example_common | ||
|
|
||
| trans_func = partial(convert_example_common, tokenizer=tokenizer, data_args=data_args) | ||
| else: | ||
| trans_func = partial(get_convert_example(model), tokenizer=tokenizer, data_args=data_args) | ||
|
|
||
| if data_args.zero_padding: |
llm/run_finetune.py
Outdated
| if dev_ds is not None | ||
| else None | ||
| ) | ||
| dev_ds = dev_ds.map(partial(trans_func)) if dev_ds is not None else None |
| rope_scaling_type: str = field(default=None, metadata={"help": "Rope extension strategy"}) | ||
| rope_scaling_factor: float = field(default=None, metadata={"help": "Rope extension scaling factor"}) | ||
| strategy_type: str = field(default=None, metadata={"help": "Long sequence strategy type"}) | ||
| strategy_name: str = field(default=None, metadata={"help": "Long sequence strategy name"}) |
There was a problem hiding this comment.
llm/utils/data.py
Outdated
|
|
||
| def get_example_pose(example, tokenizer, data_args): | ||
| if "src" in example: | ||
| source = example["src"] |
There was a problem hiding this comment.
可以和写longlora的同学https://github.com/PaddlePaddle/PaddleNLP/pull/8798/files
一起商量一下你们的数据格式到底叫src还是text，并且可以把get_example_pose写成tokenize_autogressive函数里一个分支if pose：return get_example_pose（）
|
|
|
|
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.
PR types
PR changes
Description
Pose + Yarn Algorithm