huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-10 20:04:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

Supporting code llama.

Browse Source
This commit is contained in:
Nicolas Patry 2023-08-24 18:52:20 +02:00
parent c4422e5678
commit 3db59bfd00
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_llama_modeling.py
View File

@ -64,6 +64,7 @@ class LlamaConfig(PretrainedConfig):
pretraining_tp=1,
tie_word_embeddings=False,
rope_scaling=None,
rope_theta=10000.0,
**kwargs,
):
self.vocab_size = vocab_size
@ -84,6 +85,7 @@ class LlamaConfig(PretrainedConfig):
self.pretraining_tp = pretraining_tp
self.use_cache = use_cache
self.rope_scaling = rope_scaling
self.rope_theta = rope_theta
super().__init__(
pad_token_id=pad_token_id,
@ -189,7 +191,7 @@ class FlashLlamaAttention(torch.nn.Module):
# config=config, prefix=f"{prefix}.rotary_emb", weights=weights
# )
self.rotary_emb = PositionRotaryEmbedding.static(
config=config, dim=self.head_size, base=10000.0, device=weights.device
config=config, dim=self.head_size, base=config.rope_theta, device=weights.device
)
self.softmax_scale = self.head_size**-0.5