mirrors/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-04-30 01:41:43 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 6

fix(compression): retry summary on main model for unknown errors before giving up (#16774)

Browse source 
The existing retry-on-main path in _generate_summary only fires for errors that match the _is_model_not_found heuristic (404/503, 'model_not_found', 'does not exist', 'no available channel'). Other misconfiguration errors — 400s from aggregators, provider-specific 'no route' strings, opaque rejections — fall straight through to the transient-cooldown branch, which drops N turns of context and inserts a static placeholder.

Losing context is almost always worse than one extra summary attempt. Add a best-effort retry-on-main for the unknown-error branch, guarded by the same invariants as the existing fast-path retry: only when summary_model differs from main, and only once per compressor (_summary_model_fallen_back).

Tests cover: 404 fast-path fallback still works, unknown 400 now falls back, same-model aux skips retry (no infinite loop), and a double-failure (aux + main) stops at 2 calls.
This commit is contained in:
Teknium 2026-04-27 19:25:57 -07:00 • committed by GitHub
parent f2fcc087f7
commit 94b26f3ec9
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 149 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

24
agent/context_compressor.py
View file

@ -911,6 +911,30 @@ The user has requested that this compaction PRIORITISE preserving all informatio
self._summary_failure_cooldown_until = 0.0 # no cooldown
return self._generate_summary(turns_to_summarize, focus_topic=focus_topic) # retry immediately
# Unknown-error best-effort retry on main model. Losing N turns of
# context is almost always worse than one extra summary attempt, so
# if we haven't already fallen back and the summary model differs
# from the main model, try once more on main before entering
# cooldown. Errors that DID match _is_model_not_found above are
# already handled by the fast-path retry; this branch catches
# everything else (400s, provider-specific "no route" strings,
# aggregator rejections, etc.) where auto-retry is still safer
# than dropping the turns.
if (
self.summary_model
and self.summary_model != self.model
and not getattr(self, "_summary_model_fallen_back", False)
):
self._summary_model_fallen_back = True
logging.warning(
"Summary model '%s' failed (%s). "
"Retrying on main model '%s' before giving up.",
self.summary_model, e, self.model,
)
self.summary_model = "" # empty = use main model
self._summary_failure_cooldown_until = 0.0
return self._generate_summary(turns_to_summarize, focus_topic=focus_topic)
# Transient errors (timeout, rate limit, network) — shorter cooldown
_transient_cooldown = 60
self._summary_failure_cooldown_until = time.monotonic() + _transient_cooldown