url/colibri
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #402 from anrasi/fix/coli-run-think

Browse Source 
coli: honor THINK=1 in run mode
This commit is contained in:
Vincenzo Fornaro
2026-07-19 10:39:50 +02:00
committed by GitHub
parent 6690c4f58e 799fe41f88
commit 36d389bf50
1 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
c/coli
+5 -2
View File
@@ -482,8 +482,11 @@ def cmd_run(a):
need_model(a.model) need_model(a.model)
prompt=" ".join(a.prompt) if a.prompt else sys.exit('usage: coli run "your prompt"') prompt=" ".join(a.prompt) if a.prompt else sys.exit('usage: coli run "your prompt"')
banner("run") banner("run")
# template ufficiale GLM-5.2: niente \n dopo i ruoli; <think></think> = risposta diretta (nothink) # template ufficiale GLM-5.2: niente \n dopo i ruoli; <think></think> = risposta diretta (nothink).
e=env_for(a); e["PROMPT"]=f"[gMASK]<sop><|user|>{prompt}<|assistant|><think></think>" # THINK=1 lascia <think> aperto, stessa convenzione del serve mode (glm.c). EN: THINK=1 leaves
# <think> open so the engine emits its reasoning block; the default stays nothink.
tk="<think>" if os.environ.get("THINK","0")=="1" else "<think></think>"
e=env_for(a); e["PROMPT"]=f"[gMASK]<sop><|user|>{prompt}<|assistant|>{tk}"
sys.exit(subprocess.call([GLM, str(a.cap)], env=e)) sys.exit(subprocess.call([GLM, str(a.cap)], env=e))
def server_probe(base, api_key=None, timeout=1.5): def server_probe(base, api_key=None, timeout=1.5):