deeme commited on
Commit
59f800d
·
verified ·
1 Parent(s): cd5ccc3

Upload free_ask_internet.py

Browse files
Files changed (1) hide show
  1. free_ask_internet.py +16 -23
free_ask_internet.py CHANGED
@@ -121,7 +121,8 @@ def gen_prompt(question, content_list, lang="zh-CN", context_length_limit=11000,
121
  }
122
  if len(ref_content) > 0:
123
  prompts = '''
124
- You are a large language AI assistant. You are given a user question, and please write clean, concise and accurate answer to the question. You will be given a set of related contexts to the question, each starting with a reference symbol.
 
125
 
126
  IMPORTANT INSTRUCTIONS FOR CITATIONS:
127
  1. Use ONLY the exact reference symbols provided (like ¹, ², ³) to cite sources
@@ -130,9 +131,8 @@ def gen_prompt(question, content_list, lang="zh-CN", context_length_limit=11000,
130
  4. Each citation should correspond to exactly one source
131
  5. Do not use citation symbols that weren't defined in the context
132
 
133
- Your answer must be correct, accurate and written by an expert using an unbiased and professional tone. Please limit to 1024 tokens. Do not give any information that is not related to the question, and do not repeat. Say "information is missing on" followed by the related topic, if the given context do not provide sufficient information.
134
-
135
- Here are the set of contexts:
136
  ''' + "\n\n" + "```"
137
 
138
  # 使用 Unicode 上标数字作为引用标记
@@ -275,22 +275,15 @@ def ask_internet(query:str, model:str, debug=False):
275
  else:
276
  yield response_content
277
 
278
- # 参考资料部分
279
- yield "\n\n---\n"
280
- yield "参考资料:\n"
281
- if content_list:
282
- for i, url_content in enumerate(content_list):
283
- if i >= 10: # 限制最多10个引用
284
- break
285
- url = url_content.get('url')
286
- if url:
287
- # 提取域名作为显示名称
288
- domain = tldextract.extract(url).domain
289
- subdomain = tldextract.extract(url).subdomain
290
- site_name = subdomain + "." + domain if subdomain else domain
291
-
292
- # 使用上标数字作为引用编号
293
- num = str(i + 1)
294
- sup_num = ''.join(superscript_digits.get(c, c) for c in num)
295
-
296
- yield f"{sup_num} [{site_name}]({url})\n"
 
121
  }
122
  if len(ref_content) > 0:
123
  prompts = '''
124
+ You are a large language AI assistant. You are given a user question, and please write clean, concise and accurate answer to the question. You will be given a set of related contexts to the question, each starting with a reference symbol. Please use the context and cite the context at the end of each sentence if applicable.
125
+ Your answer must be correct, accurate and written by an expert using an unbiased and professional tone. Please limit to 1024 tokens. Do not give any information that is not related to the question, and do not repeat. Say "information is missing on" followed by the related topic, if the given context do not provide sufficient information.
126
 
127
  IMPORTANT INSTRUCTIONS FOR CITATIONS:
128
  1. Use ONLY the exact reference symbols provided (like ¹, ², ³) to cite sources
 
131
  4. Each citation should correspond to exactly one source
132
  5. Do not use citation symbols that weren't defined in the context
133
 
134
+ Please cite the contexts with the reference numbers, in the format (like ¹, ², ³) . If a sentence comes from multiple contexts, please list all applicable citations, like(¹ ²). Other than code and specific names and citations, your answer must be written in the same language as the question.
135
+ Here are the set of contexts:
 
136
  ''' + "\n\n" + "```"
137
 
138
  # 使用 Unicode 上标数字作为引用标记
 
275
  else:
276
  yield response_content
277
 
278
+ yield "\n\n"
279
+ # 是否返回参考资料
280
+ if True:
281
+ yield "---"
282
+ yield "\nSearxng"
283
+ yield "参考资料:\n"
284
+ count = 1
285
+ for url_content in content_list:
286
+ url = url_content.get('url')
287
+ yield "*[{}. {}]({})*".format(str(count),url,url )
288
+ yield "\n"
289
+ count += 1