huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 12:24:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: default use_qk_norm false in cohere

Browse Source
This commit is contained in:
drbh 2024-04-17 20:59:16 +00:00
parent 06c3d4b1ec
commit 91c653bac2
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
server/text_generation_server/models/custom_modeling/flash_cohere_modeling.py
View File

@ -216,7 +216,9 @@ class FlashCohereAttention(torch.nn.Module):
self.query_key_value = load_attention(config, prefix, weights) self.query_key_value = load_attention(config, prefix, weights)
self.use_qk_norm = config.use_qk_norm self.use_qk_norm = (
config.use_qk_norm if hasattr(config, "use_qk_norm") else False
)
if self.use_qk_norm: if self.use_qk_norm:
self.q_norm = CohereLayerNorm( self.q_norm = CohereLayerNorm(
prefix=f"{prefix}.q_norm", prefix=f"{prefix}.q_norm",