inference support llama3(wint8|4/a8w8) #8630
Merged
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! |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## develop #8630 +/- ##
========================================
Coverage 55.80% 55.80%
========================================
Files 620 620
Lines 96642 96642
========================================
Hits 53928 53928
Misses 42714 42714 ☔ View full report in Codecov by Sentry. |
1 task
yuanlehome
force-pushed
the
support_llama3_inferernce
branch
from
e40bdf1
to
a81c16d
Compare
DesmonDay
reviewed
Jun 25, 2024
llm/predict/predictor.py
Outdated
| @@ -1213,8 +1214,8 @@ def create_predictor( | |||
| init_chat_template(tokenizer, predictor_args.model_name_or_path, predictor_args.chat_template) | |||
|
|
|||
| # TODO(wj-Mcat): fix llama tokenzier pad_token bug | |||
| if isinstance(tokenizer, LlamaTokenizer) and not tokenizer.pad_token: | |||
| tokenizer.pad_token = tokenizer.unk_token | |||
| if (isinstance(tokenizer, LlamaTokenizer) or isinstance(tokenizer, Llama3Tokenizer)) and not tokenizer.pad_token: | |||
There was a problem hiding this comment.
这块地方可以简化为：if (isinstance(tokenizer, (LlamaTokenizer, Llama3Tokenizer)) and not tokenizer.pad_token:。isintance支持元组输入。
DesmonDay
reviewed
Jun 25, 2024
| @@ -549,7 +549,7 @@ def init_weight_shape(self, config): | |||
| self.qkv_weight_shape = ( | |||
| [(self.num_heads + 2 * self.kv_num_heads) * self.head_dim, self.embed_dim] | |||
| if config.trans_qkvw | |||
| else [(self.num_heads + 2 * self.kv_num_heads) * self.head_dim, self.embed_dim] | |||
| else [self.embed_dim, (self.num_heads + 2 * self.kv_num_heads) * self.head_dim] | |||
yuanlehome
force-pushed
the
support_llama3_inferernce
branch
from
833a8a7
to
7130c18
Compare
|
如讨论，目前llama3模型，在动态图非fuse场景下推理正常，在fuse场景下推理存在多进程问题。待后续排查。另外动转静时不可以设置src_length进行推理，以及高性能推理下无法正确eos。 @yuanlehome |
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
New features
PR changes
Others
Description
inference support llama3(wint8|4/a8w8)