fix(distill): max_tokens 800→1200 — Agnes推理token预算不足

Agnes 2.0-flash是推理模型,推理token占大量预算。
800仍不够→JSON截断(parse error: unexpected end of JSON input)。
1200确保推理+正文都够用。
This commit is contained in:
xiaowei 2026-08-20 01:13:06 +08:00
parent ea4b5a4dfe
commit ae73982743
3 changed files with 2 additions and 2 deletions

View File

@ -184,7 +184,7 @@ func (e *Engine) ConsolidateMemory(target MemoryCandidate, candidates []MemoryCa
"model": e.LLMModel,
"messages": []map[string]string{{"role": "user", "content": prompt}},
"temperature": 0.1,
"max_tokens": 800,
"max_tokens": 1200,
}
jsonBody, err := json.Marshal(body)
if err != nil {

View File

@ -339,7 +339,7 @@ func (e *Engine) callLLM5D(content string) (LLMResponse, error) {
{"role": "user", "content": prompt},
},
"temperature": 0.2,
"max_tokens": 800,
"max_tokens": 1200,
}
jsonBody, err := json.Marshal(body)

Binary file not shown.