Update README.md

README.md

@@ -24,6 +24,8 @@ pip install auto-gptq
 Since the original auto-gptq library does not support the quantization of chatglm models, manual import (hack) is required.
 ```python
 from auto_gptq.modeling._base import BaseGPTQForCausalLM
+from auto_gptq.modeling._const import SUPPORTED_MODELS
+
 from auto_gptq.modeling.auto import GPTQ_CAUSAL_LM_MODEL_MAP
 
 class ChatGLMGPTQForCausalLM(BaseGPTQForCausalLM):
@@ -42,6 +44,7 @@ class ChatGLMGPTQForCausalLM(BaseGPTQForCausalLM):
     ]
 
 GPTQ_CAUSAL_LM_MODEL_MAP['chatglm'] = ChatGLMGPTQForCausalLM
+SUPPORTED_MODELS = SUPPORTED_MODELS.append('chatglm')
 ```
 
 The complete model import code is as follows:
@@ -61,6 +64,8 @@ from auto_gptq import AutoGPTQForCausalLM
 
 
 from auto_gptq.modeling._base import BaseGPTQForCausalLM
+from auto_gptq.modeling._const import SUPPORTED_MODELS
+
 from auto_gptq.modeling.auto import GPTQ_CAUSAL_LM_MODEL_MAP
 
 class ChatGLMGPTQForCausalLM(BaseGPTQForCausalLM):
@@ -79,6 +84,7 @@ class ChatGLMGPTQForCausalLM(BaseGPTQForCausalLM):
     ]
 
 GPTQ_CAUSAL_LM_MODEL_MAP['chatglm'] = ChatGLMGPTQForCausalLM
+SUPPORTED_MODELS = SUPPORTED_MODELS.append('chatglm')
 
 device = 'cuda:0'
 quantized_model_dir = 'alexwww94/glm-4v-9b-gptq'