[Bugfix] Fix prefix strings for quantized VLMs (#9772)

This commit is contained in:
Michael Goin
2024-10-29 16:02:59 -07:00
committed by GitHub
parent 8d7724104a
commit bc73e9821c
20 changed files with 288 additions and 97 deletions
+4 -1
View File
@@ -439,7 +439,10 @@ class InternVLChatModel(nn.Module, SupportsMultiModal, SupportsPP):
)
self.language_model = init_vllm_registered_model(
config.text_config, cache_config, quant_config)
config.text_config,
cache_config,
quant_config,
prefix="language_model")
self.mlp1 = self._init_mlp1(config)