qwen2

qwen2

Qwen2 and Qwen2.5 on vLLM (vllm.model_executor.models.qwen2).

Llama's names and Llama's fused block, in classes of its own.

RENAME module-attribute

RENAME = {'model.embed_tokens': 'embed_tokens', 'model.layers': 'layers', 'model.norm': 'norm'}

ENVOYS module-attribute

ENVOYS = {Qwen2DecoderLayer: Layer, Qwen2Attention: Attention, Qwen2MLP: Mlp}

Layer

Layer(*args: Any, **kwargs: Any)

Bases: FusedLayer

Qwen2's decoder block: forward(positions, hidden_states, residual) -> (hidden_states, residual).