Aibo117 fa9f64752b fix: reduce chunk size to prevent embedding context overflow ...

nomic-embed-text tokenizes Chinese characters differently from the
estimateTokenCount heuristic (3-5 real tokens per character vs 1
estimated). Large chunks overflow the 8192 token context. Reduce
TargetTokens/MaxTokens to keep chunks within limits.

2026-05-26 14:56:24 +08:00

..

application/runtime

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00

mcps

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00

rag

fix: reduce chunk size to prevent embedding context overflow

2026-05-26 14:56:24 +08:00

runtime

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00

skills

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00

embedding.go

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00

llm.go

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00

openai_client.go

refactor: rebrand from cs-agent to BossChat

2026-05-26 13:17:07 +08:00