diff options
| author | Danilo M. <danix@danix.xyz> | 2026-08-11 11:22:32 +0200 |
|---|---|---|
| committer | Danilo M. <danix@danix.xyz> | 2026-08-11 11:22:32 +0200 |
| commit | 2cae7f19ac90e6df9d4008e6e364bae30e8b389a (patch) | |
| tree | 045eae84caefd1ebd97836fe0a5b97ae181022a2 /llamachat.py | |
| parent | 3f984ccb8ec25ef3b06109d638eed4022db82d45 (diff) | |
| download | llamachat-2cae7f19ac90e6df9d4008e6e364bae30e8b389a.tar.gz llamachat-2cae7f19ac90e6df9d4008e6e364bae30e8b389a.zip | |
fix: prevent cloud reasoning models from exhausting default max_tokens
- Send max_tokens = 32768 for non-local providers so reasoning models have room for both chain-of-thought and answer.
- Add per-provider thinking_budget option for endpoints (currently SiliconFlow) that cap reasoning tokens separately.
- Accept bare JSON arrays from /v1/models; some OpenAI-compatible endpoints omit the {"data": [...]} envelope.
Diffstat (limited to 'llamachat.py')
0 files changed, 0 insertions, 0 deletions