Peng Xu
Results
2
comments of
Peng Xu
fwiw, on vllm-0.7.2, running the same snippet: ``` from vllm import LLM llm = LLM(model="Skywork/Skywork-Reward-Llama-3.1-8B-v0.2", task="reward") (output,) = llm.encode("Hello, my name is") data = output.outputs.data print(f"Data: {data!r}") ``` I got...
Hi @DarkLight1337 I think `llm.encode()` is just pulling the hidden states of the last layer even we set `task="reward"`. Here is an example snippet that replicates the issue ``` from...