apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 08:37:06 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[FDConfig]Remove max_num_batched_tokens/max_num_seqs in parallel config (#4116)

Browse Source 
* remove max_num_batched_tokens in parallel config

* remove max_num_seqs

* update test case

* fix test

* fix

---------

Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com>
This commit is contained in:
YuanRisheng
2025-09-17 10:43:35 +08:00
committed by GitHub
parent c01a756912
commit 2e9e53ff7e
30 changed files with 169 additions and 131 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
fastdeploy/model_executor/models/deepseek_v3.py
View File

@@ -607,9 +607,11 @@ class DeepseekV3ForCausalLM(ModelForCasualLM):
num_embeddings=fd_config.model_config.vocab_size,
prefix="lm_head",
)
self.position_ids_buffer = paddle.empty([fd_config.parallel_config.max_num_batched_tokens], dtype=paddle.int32)
self.position_ids_buffer = paddle.empty(
[fd_config.scheduler_config.max_num_batched_tokens], dtype=paddle.int32
)
self.mask_encoder_batch_buffer = paddle.empty(
[fd_config.parallel_config.max_num_batched_tokens, 1], dtype=paddle.int32
[fd_config.scheduler_config.max_num_batched_tokens, 1], dtype=paddle.int32
)
@classmethod