Hermes configuration contexte en dur ...

[Tuxedo17]

Voici la configuration :

model:
  default: llama-swap
  provider: custom
  base_url: http://127.0.0.1:8080/v1
  api_key: llama-swap
  context_length: 175000

Via llama-swap quand je configure à 180.000 j’ai 131.000 . Je configure donc en dur 175000.

[Tuxedo17]

J’ai quand même des compressions quand je suis à 131.000 …

[Tuxedo17]

Ma configuration :

compression:
  enabled: true
  threshold: 0.5
  target_ratio: 0.2
  protect_last_n: 20
  protect_first_n: 3
  codex_gpt55_autoraise: true
  codex_app_server_auto: native
...
memory:
  memory_enabled: true
  user_profile_enabled: true
  memory_char_limit: 2200
  user_char_limit: 1375
  nudge_interval: 10
  flush_min_turns: 6

[Tuxedo17]

Requete pour la compression :

{
  "messages": [
    {
      "role": "user",
      "content": "You are a summarization agent creating a context checkpoint. Treat the conversation turns below as source material for a compact record of prior work. Produce only the structured summary; do not add a greeting, preamble, or prefix. Write the summary in the same language the user was using in the conversation — do not translate or switch to English. NEVER include API keys, tokens, passwords, secrets, credentials, or connection strings in the summary — replace any that appear with [REDACTED]. Note that the user had credentials present, but do not preserve their values.\n\nCreate a structured checkpoint summary for the conversation after earlier turns are compacted. The summary should preserve enough detail for continuity without re-reading the original turns