Folzo/Effi
Public Access
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

jean-router: utiliser le nom d'affichage (name) comme id modele

Browse Source 
- /v1/models expose name au lieu de l'id technique des presets Jean
- Routing: resolution name -> id interne pour le switch
- Correspondance bidirectionnelle dans chat_completions
This commit is contained in:
Folzo
2026-07-26 21:30:32 +02:00
parent cbb43ad0f3
commit 68a76aa254
1 changed files with 14 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
jean-router/app.py
+14 -9
View File
@@ -554,6 +554,7 @@ def list_models():
for p in presets: for p in presets:
model_id = p["id"] model_id = p["id"]
model_name = p.get("name", model_id)
ctx_len = jean_ctx.get(model_id) ctx_len = jean_ctx.get(model_id)
caps = jean_caps.get(model_id, ["completion", "tools"]) caps = jean_caps.get(model_id, ["completion", "tools"])
@@ -561,12 +562,12 @@ def list_models():
caps = list(dict.fromkeys(caps + ["thinking"])) caps = list(dict.fromkeys(caps + ["thinking"]))
model_info = { model_info = {
"id": model_id, "id": model_name,
"object": "model", "object": "model",
"created": int(time.time()), "created": int(time.time()),
"owned_by": "jean", "owned_by": "jean",
"permission": [], "permission": [],
"root": model_id, "root": model_name,
"parent": None, "parent": None,
} }
@@ -659,26 +660,30 @@ def chat_completions():
return jsonify({"error": "model parameter required"}), 400 return jsonify({"error": "model parameter required"}), 400
presets = get_presets() presets = get_presets()
jean_preset_ids = [p["id"] for p in presets] jean_preset_ids = {p["id"] for p in presets}
jean_name_to_id = {p.get("name", p["id"]): p["id"] for p in presets}
jean_names = set(jean_name_to_id.keys())
if model_id in jean_preset_ids or model_id in jean_names:
internal_id = jean_name_to_id.get(model_id, model_id)
if model_id in jean_preset_ids:
if not ensure_vram_for_backend("jean"): if not ensure_vram_for_backend("jean"):
return jsonify({"error": "Failed to prepare VRAM for Jean"}), 503 return jsonify({"error": "Failed to prepare VRAM for Jean"}), 503
with switch_lock: with switch_lock:
active = get_active_preset() active = get_active_preset()
current_active = active["id"] if active else last_active_preset current_active = active["id"] if active else last_active_preset
if current_active != model_id: if current_active != internal_id:
log.info(f"[Router] Switch Jean requis: {current_active or 'none'} → {model_id}") log.info(f"[Router] Switch Jean requis: {current_active or 'none'} → {internal_id} ({model_id})")
if not switch_preset(model_id): if not switch_preset(internal_id):
return jsonify({"error": f"Failed to switch to preset {model_id}"}), 503 return jsonify({"error": f"Failed to switch to preset {model_id}"}), 503
else: else:
log.info(f"[Router] Preset {model_id} déjà actif (tracké)") log.info(f"[Router] Preset {model_id} déjà actif (tracké)")
preset = next((p for p in presets if p["id"] == model_id), {}) preset = next((p for p in presets if p["id"] == internal_id), {})
if preset.get("reasoning") == "on": if preset.get("reasoning") == "on":
ctx, _ = get_jean_metadata() ctx, _ = get_jean_metadata()
ctx_len = ctx.get(model_id, 131072) ctx_len = ctx.get(internal_id, 131072)
req_max = data.get("max_tokens", 0) req_max = data.get("max_tokens", 0)
if req_max and req_max > 0: if req_max and req_max > 0:
new_max = min(req_max * 2, ctx_len) new_max = min(req_max * 2, ctx_len)