Skip to content

GitLab

Unverified Commit d56afd45 authored by Andy Lo's avatar Andy Lo Committed by GitHub
Browse files

Remove unused logic in `models/mistral.py` (#33095) 


Signed-off-by: default avatarAndy Lo <andy@mistral.ai>
parent a2393ed4
Show whitespace changes
Inline Side-by-side
Showing with 0 additions and 8 deletions
vllm/model_executor/models/mistral.py
View file @ d56afd45
...@@ -156,16 +156,8 @@ class MistralDecoderLayer(LlamaDecoderLayer): ...@@ -156,16 +156,8 @@ class MistralDecoderLayer(LlamaDecoderLayer):
) )
self.layer_idx = int(prefix.split(sep=".")[-1]) self.layer_idx = int(prefix.split(sep=".")[-1])
quant_config = self.get_quant_config(vllm_config)
config = config or vllm_config.model_config.hf_config config = config or vllm_config.model_config.hf_config
do_fusion = getattr(
quant_config, "enable_quantization_scaling_fusion", False
) and vllm_config.cache_config.cache_dtype.startswith("fp8")
if do_fusion:
self.input_layernorm.quant_scaling_from = self.self_attn.qkv_proj
self.post_attention_layernorm.quant_scaling_from = self.mlp.gate_up_proj
if getattr(config, "ada_rms_norm_t_cond", False): if getattr(config, "ada_rms_norm_t_cond", False):
self.ada_rms_norm_t_cond = nn.Sequential( self.ada_rms_norm_t_cond = nn.Sequential(
ColumnParallelLinear( ColumnParallelLinear(
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment