EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[QUANT] Add GPTQModel Dynamic Quantization + lm_head Quantization (#3790)

Browse Source 
Signed-off-by: ZX-ModelCloud <zx@modelcloud.ai>
Co-authored-by: ZX-ModelCloud <zx@modelcloud.ai>
This commit is contained in:
Qubitium-ModelCloud
2025-03-05 17:11:00 +08:00
committed by GitHub
parent 583d6af71b
commit 56a724eba3
56 changed files with 1988 additions and 282 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
python/sglang/srt/models/llama_classification.py
View File

@@ -23,6 +23,7 @@ from sglang.srt.layers.quantization.base_config import QuantizationConfig
from sglang.srt.model_executor.forward_batch_info import ForwardBatch
from sglang.srt.model_loader.weight_utils import default_weight_loader
from sglang.srt.models.llama import LlamaForCausalLM, LlamaModel
from sglang.srt.utils import add_prefix
class LlamaForClassification(nn.Module):
@@ -30,11 +31,14 @@ class LlamaForClassification(nn.Module):
self,
config: LlamaConfig,
quant_config: Optional[QuantizationConfig] = None,
prefix: str = "",
) -> None:
super().__init__()
self.config = config
self.quant_config = quant_config
self.model = LlamaModel(config, quant_config=quant_config)
self.model = LlamaModel(
config, quant_config=quant_config, prefix=add_prefix("model", prefix)
)
self.classification_head = nn.Linear(
config.hidden_size, config.classification_out_size, bias=False