Skip to content

vllm.model_executor.models.ernie45

Inference-only Erine model compatible with HuggingFace weights.

Ernie4_5_ForCausalLM

Bases: LlamaForCausalLM

Source code in vllm/model_executor/models/ernie45.py 
31
32
33
34
35
36
37
38
39
40
41
42
43
class Ernie4_5_ForCausalLM(LlamaForCausalLM):

    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
        super().__init__(vllm_config=vllm_config, prefix=prefix)
        # Hack Llama model to fit HF format Ernie4.5 dense implementation
        # Attention difference between Ernie and Llama:
        # 1. rotary_dim and no Neox style.
        # 2. There is no bias for o_proj in attention
        for layer in self.model.layers:
            if not isinstance(layer, PPMissingLayer):
                layer.self_attn.rotary_emb.is_neox_style = False
                layer.self_attn.o_proj.bias = None
                layer.self_attn.o_proj.skip_bias_add = True

__init__ 

__init__(*, vllm_config: VllmConfig, prefix: str = '')
Source code in vllm/model_executor/models/ernie45.py 
33
34
35
36
37
38
39
40
41
42
43
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
    super().__init__(vllm_config=vllm_config, prefix=prefix)
    # Hack Llama model to fit HF format Ernie4.5 dense implementation
    # Attention difference between Ernie and Llama:
    # 1. rotary_dim and no Neox style.
    # 2. There is no bias for o_proj in attention
    for layer in self.model.layers:
        if not isinstance(layer, PPMissingLayer):
            layer.self_attn.rotary_emb.is_neox_style = False
            layer.self_attn.o_proj.bias = None
            layer.self_attn.o_proj.skip_bias_add = True