ollama - 💡(How to fix) Fix Ollama Claude Code context auto-compact/Timeout issue [1 comments, 2 participants]

TL;DR

• Verify that the Context Length set in Ollama is correctly propagated to Claude and adjust the GPU size to fit the full context size if possible.

Guidance

• Check the Ollama configuration to ensure that the Context Length is correctly set and matches the expected value.
• Investigate why Claude is not performing AutoCompact or warning when the context length is reached, potentially due to a mismatch between the set context length and the model's max token limit.
• Review the GPU size and model requirements to determine if the full context size can fit within the available GPU memory.
• Compare the output of the /context command with the expected output to identify any discrepancies in token usage reporting.

Example

• No code snippet is provided as the issue lacks specific technical details about the implementation.

Notes

• The issue may be related to a configuration mismatch or a limitation in the current Ollama version (0.18.2).
• The provided screenshot may contain relevant information, but its content is not accessible in this format.

Recommendation

• Apply workaround: Adjust the GPU size to fit the full context size if possible, and verify that the Context Length set in Ollama is correctly propagated to Claude, as this may help mitigate the issue until a more permanent fix is available.