Skip to content

fix(agent): preserve global model.context_length through switch_model and fallback #32413

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
TF0rd wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix(agent): preserve global model.context_length through switch_model and fallback #32413

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
30 changes: 17 additions & 13 deletions agent/agent_runtime_helpers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1361,9 +1361,23 @@ def switch_model(agent, new_model, new_provider, api_key='', base_url='', api_mo
old_model = agent.model
old_provider = agent.provider

# Clear the per-config context_length override so the new model's
# Clear the per-model _config_context_length override so the new model's
# actual context window is resolved via get_model_context_length()
# instead of inheriting the stale value from the previous model.
# instead of inheriting a stale per-model value from the previous model.
# However, preserve the GLOBAL model.context_length from config.yaml
# — that setting applies to ALL models and must survive switches.
_global_config_ctx = None
_sm_cfg = None
_sm_custom_providers = None
try:
from hermes_cli.config import load_config as _sm_load_cfg, get_compatible_custom_providers
_sm_cfg = _sm_load_cfg()
_sm_model_cfg = _sm_cfg.get("model", {})
if isinstance(_sm_model_cfg, dict) and _sm_model_cfg.get("context_length") is not None:
_global_config_ctx = int(_sm_model_cfg["context_length"])
_sm_custom_providers = get_compatible_custom_providers(_sm_cfg)
except Exception:
pass
agent._config_context_length = None

# ── Swap core runtime fields ──
Expand Down Expand Up @@ -1453,16 +1467,6 @@ def switch_model(agent, new_model, new_provider, api_key='', base_url='', api_mo
# ── Update context compressor ──
if hasattr(agent, "context_compressor") and agent.context_compressor:
from agent.model_metadata import get_model_context_length
# Re-read custom_providers from live config so per-model
# context_length overrides are honored when switching to a
# custom provider mid-session (closes #15779).
_sm_custom_providers = None
try:
from hermes_cli.config import load_config, get_compatible_custom_providers
_sm_cfg = load_config()
_sm_custom_providers = get_compatible_custom_providers(_sm_cfg)
except Exception:
_sm_custom_providers = None
# ``agent.api_key`` may be a callable (Azure Foundry Entra ID
# token provider). ``get_model_context_length`` expects a
# string for its live-probe paths; for Foundry the context
Expand All @@ -1474,7 +1478,7 @@ def switch_model(agent, new_model, new_provider, api_key='', base_url='', api_mo
base_url=agent.base_url,
api_key=_ctx_api_key,
provider=agent.provider,
config_context_length=getattr(agent, "_config_context_length", None),
config_context_length=_global_config_ctx,
custom_providers=_sm_custom_providers,
)
agent.context_compressor.update_model(
Expand Down
12 changes: 11 additions & 1 deletion agent/chat_completion_helpers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1010,6 +1010,16 @@ def try_activate_fallback(agent, reason: "FailoverReason | None" = None) -> bool
# Clear the per-config context_length override so the fallback
# model's actual context window is resolved instead of inheriting
# the stale value from the previous model. See #22387.
# Preserve the GLOBAL model.context_length from config.yaml —
# that setting applies to ALL models and must survive fallbacks.
_global_config_ctx = None
try:
from hermes_cli.config import load_config as _fb_load_cfg
_fb_model_cfg = _fb_load_cfg().get("model", {})
if isinstance(_fb_model_cfg, dict) and _fb_model_cfg.get("context_length") is not None:
_global_config_ctx = int(_fb_model_cfg["context_length"])
except Exception:
pass
agent._config_context_length = None
agent.model = fb_model
agent.provider = fb_provider
Expand Down Expand Up @@ -1094,7 +1104,7 @@ def try_activate_fallback(agent, reason: "FailoverReason | None" = None) -> bool
fb_context_length = get_model_context_length(
agent.model, base_url=agent.base_url,
api_key=_fb_ctx_api_key, provider=agent.provider,
config_context_length=getattr(agent, "_config_context_length", None),
config_context_length=_global_config_ctx,
custom_providers=getattr(agent, "_custom_providers", None),
)
agent.context_compressor.update_model(
Expand Down