@@ -320,7 +320,7 @@ def prepare_tensors(self):
320320 gguf .LlamaFileType .MOSTLY_Q4_0 ,
321321 gguf .LlamaFileType .MOSTLY_Q4_1 ,
322322 ):
323- data_qtype = gguf .GGMLQuantizationType .Q5_0
323+ data_qtype = gguf .GGMLQuantizationType .Q5_0
324324 elif self .ftype in (
325325 gguf .LlamaFileType .MOSTLY_Q5_0 ,
326326 gguf .LlamaFileType .MOSTLY_Q5_1 ,
@@ -418,7 +418,7 @@ def prepare_metadata(self, vocab_only: bool):
418418
419419 logger .info ("Set model quantization version" )
420420 self .gguf_writer .add_quantization_version (gguf .GGML_QUANT_VERSION )
421-
421+
422422 logger .info ("***********************************************************************************************" )
423423 logger .info ("** Converting to `q4_0`,`q4_1`,`q5_0`, `q5_1` or `q6_0` is not equiv to using `llama-quantize`!" )
424424 logger .info ("** Ftype `q4_0`,`q4_1` are here converting embeddings, output, attn_k and attn_v/qkv in q5_0." )
@@ -2195,8 +2195,8 @@ class Qwen3MoeModel(Qwen2MoeModel):
21952195 model_arch = gguf .MODEL_ARCH .QWEN3MOE
21962196
21972197
2198- @ModelBase .register ("Ernie4_5_ForCausalLM" , "Ernie4_5ForCausalLM" )
2199- class Ernie4_5Model (TextModel ):
2198+ @Model .register ("Ernie4_5_ForCausalLM" , "Ernie4_5ForCausalLM" )
2199+ class Ernie4_5Model (Model ):
22002200 model_arch = gguf .MODEL_ARCH .ERNIE4_5
22012201
22022202 def set_vocab (self ):
@@ -2242,7 +2242,7 @@ def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iter
22422242 return [(self .map_tensor_name (name ), data_torch )]
22432243
22442244
2245- @ModelBase .register ("Ernie4_5_MoeForCausalLM" )
2245+ @Model .register ("Ernie4_5_MoeForCausalLM" )
22462246class Ernie4_5MoeModel (Ernie4_5Model ):
22472247 model_arch = gguf .MODEL_ARCH .ERNIE4_5_MOE
22482248 _experts : list [dict [str , Tensor ]] | None = None
0 commit comments