camel-ai / crab

CRAB: Cross-environment Agent Benchmark for Multimodal Language Model Agents. https://crab.camel-ai.org/
https://crab.camel-ai.org/
186 stars 26 forks source link

[BUG] I am using CamelModel to run the CRAB agent on top of Groq #44

Closed Dineshkumar-Anandan-ZS0367 closed 2 weeks ago

Dineshkumar-Anandan-ZS0367 commented 1 month ago

What version of camel are you using?

0.2.1

Problem description

ValueError: Unexpected argument logit_bias is input into Groq model backend.

Reproducible example code

if name == "main": benchmark = create_benchmark(template_benchmark_config) task, action_space = benchmark.start_task("0") env_descriptions = benchmark.get_env_descriptions()

agent = SingleAgentPolicy(
    model_backend=CamelModel(
        "llama-3.1-70b-versatile",
        model_platform='groq',
        parameters={
            "max_tokens": 8192,
            "temperature": 0.6,
            "top_p": 1.0,
            "n": 1,
            "stream": False,
            "presence_penalty": 0.0,
            "frequency_penalty": 0.0,
            "tool_choice": "auto",

        }
    )
)

agent.reset(task.description, action_space, env_descriptions)
print("Start performing task: " + colored(f'"{task.description}"', "green"))
start_benchmark(benchmark, agent)
benchmark.reset()
WHALEEYE commented 4 weeks ago

Hi @Dineshkumar-Anandan-ZS0367, thanks for your feedback! We have already located the problem, and we are fixing it. This should be done by the end of next week.

WHALEEYE commented 3 weeks ago

Hi @Dineshkumar-Anandan-ZS0367, the problem is fixed in another branch and will be merged into the main branch soon through PR #45. The CAMEL integration of CRAB is currently still unstable, and your feedback is really precious to us. Feel free to reach out when you meet other problems :)

WHALEEYE commented 2 weeks ago

Hi @Dineshkumar-Anandan-ZS0367, since PR #45 has been merged we are closing this Issue. Thanks for your feedback!