| """Dependency-free Ollama inference for a fully rendered memory task prompt.""" |
| import argparse, json, urllib.request |
| import prompt, structured |
|
|
| def main(): |
| p=argparse.ArgumentParser() |
| p.add_argument('--model',default='mem-extractor') |
| p.add_argument('--host',default='http://127.0.0.1:11434') |
| p.add_argument('--prompt-file',required=True) |
| a=p.parse_args() |
| with open(a.prompt_file) as f: user=f.read() |
| user,schema,paired=structured.prepare(user) |
| if prompt.estimate_tokens(user)+3072>12288:raise ValueError('Structured prompt exceeds the tested context budget') |
| payload={'model':a.model,'system':prompt.SYSTEM,'prompt':user,'format':schema,'think':False,'stream':False,'options':{'temperature':0,'num_ctx':12288,'num_predict':3072}} |
| req=urllib.request.Request(a.host.rstrip('/')+'/api/generate',data=json.dumps(payload).encode(),headers={'Content-Type':'application/json'}) |
| with urllib.request.urlopen(req,timeout=600) as response: result=json.load(response) |
| if result.get('prompt_eval_count',12288)+16>=12288:raise RuntimeError('Prompt truncation cannot be ruled out; no canonical result returned') |
| if result.get('done_reason')=='length':raise RuntimeError('Model output was truncated; no canonical result returned') |
| print(json.dumps(structured.canonicalize(json.loads(result['response']),paired),ensure_ascii=False,indent=2)) |
|
|
| if __name__=='__main__': main() |
|
|
