Reflect fails on Bedrock Kimi K3: model rejects the temperature field

Open Beginner friendly
#6,683 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
76/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
python
Domain
ai, backend

Research direction

Start with hindsight_api/engine/reflect/agent.py to trace where the reflect temperature is set, then read engine/providers/litellm_llm.py to see how the provider passes it through. Compare with how reasoning models that reject temperature are handled. Done when reflect works with the Bedrock Kimi K3 model while temperature is unset or dropped.

Written by the indexing model from the issue text.

Description

core

What happens: every reflect on bedrock/us.moonshotai.kimi-k3 fails on main with:

BedrockException - {"message":"This model doesn't support the temperature field. Remove temperature and try again."}

How to see it: run run_reflect_agent with LLMProvider(provider="bedrock", model="bedrock/us.moonshotai.kimi-k3") (found during the smoke run for #6671; 2/2 runs failed on main, before that PR).

Expected: reflect works, with temperature left unset for this model, the way reasoning models that refuse a temperature already are.

Where: reflect sends temperature=get_config().llm_temperature_reflect (hindsight_api/engine/reflect/agent.py), and the LiteLLM provider passes it through (engine/providers/litellm_llm.py). The model probably needs to be known as one that refuses temperature, or LiteLLM's drop_params used for it.

Dominant language
Python
Stars
45.2k
Forks
5.9k
Avg merge
5h 45m
Merged PRs (30d)
355

Getting set up

This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from vectorize-io/hindsight

All issues in vectorize-io/hindsight

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.