Fix errors detected by Ruff (#3918)

### What problem does this PR solve?

Fix errors detected by Ruff

### Type of change

- [x] Refactoring
This commit is contained in:
Zhichang Yu
2024-12-08 14:21:12 +08:00
committed by GitHub
parent e267a026f3
commit 0d68a6cd1b
97 changed files with 2558 additions and 1976 deletions

View File

@ -106,15 +106,15 @@ def message_fit_in(msg, max_length=4000):
return c, msg
ll = num_tokens_from_string(msg_[0]["content"])
l = num_tokens_from_string(msg_[-1]["content"])
if ll / (ll + l) > 0.8:
ll2 = num_tokens_from_string(msg_[-1]["content"])
if ll / (ll + ll2) > 0.8:
m = msg_[0]["content"]
m = encoder.decode(encoder.encode(m)[:max_length - l])
m = encoder.decode(encoder.encode(m)[:max_length - ll2])
msg[0]["content"] = m
return max_length, msg
m = msg_[1]["content"]
m = encoder.decode(encoder.encode(m)[:max_length - l])
m = encoder.decode(encoder.encode(m)[:max_length - ll2])
msg[1]["content"] = m
return max_length, msg
@ -257,7 +257,8 @@ def chat(dialog, messages, stream=True, **kwargs):
idx = set([kbinfos["chunks"][int(i)]["doc_id"] for i in idx])
recall_docs = [
d for d in kbinfos["doc_aggs"] if d["doc_id"] in idx]
if not recall_docs: recall_docs = kbinfos["doc_aggs"]
if not recall_docs:
recall_docs = kbinfos["doc_aggs"]
kbinfos["doc_aggs"] = recall_docs
refs = deepcopy(kbinfos)
@ -433,13 +434,15 @@ def relevant(tenant_id, llm_id, question, contents: list):
Give a binary score 'yes' or 'no' score to indicate whether the document is relevant to the question.
No other words needed except 'yes' or 'no'.
"""
if not contents:return False
if not contents:
return False
contents = "Documents: \n" + " - ".join(contents)
contents = f"Question: {question}\n" + contents
if num_tokens_from_string(contents) >= chat_mdl.max_length - 4:
contents = encoder.decode(encoder.encode(contents)[:chat_mdl.max_length - 4])
ans = chat_mdl.chat(prompt, [{"role": "user", "content": contents}], {"temperature": 0.01})
if ans.lower().find("yes") >= 0: return True
if ans.lower().find("yes") >= 0:
return True
return False
@ -481,8 +484,10 @@ Requirements:
]
_, msg = message_fit_in(msg, chat_mdl.max_length)
kwd = chat_mdl.chat(prompt, msg[1:], {"temperature": 0.2})
if isinstance(kwd, tuple): kwd = kwd[0]
if kwd.find("**ERROR**") >=0: return ""
if isinstance(kwd, tuple):
kwd = kwd[0]
if kwd.find("**ERROR**") >=0:
return ""
return kwd
@ -508,8 +513,10 @@ Requirements:
]
_, msg = message_fit_in(msg, chat_mdl.max_length)
kwd = chat_mdl.chat(prompt, msg[1:], {"temperature": 0.2})
if isinstance(kwd, tuple): kwd = kwd[0]
if kwd.find("**ERROR**") >= 0: return ""
if isinstance(kwd, tuple):
kwd = kwd[0]
if kwd.find("**ERROR**") >= 0:
return ""
return kwd
@ -520,7 +527,8 @@ def full_question(tenant_id, llm_id, messages):
chat_mdl = LLMBundle(tenant_id, LLMType.CHAT, llm_id)
conv = []
for m in messages:
if m["role"] not in ["user", "assistant"]: continue
if m["role"] not in ["user", "assistant"]:
continue
conv.append("{}: {}".format(m["role"].upper(), m["content"]))
conv = "\n".join(conv)
today = datetime.date.today().isoformat()
@ -581,7 +589,8 @@ Output: What's the weather in Rochester on {tomorrow}?
def tts(tts_mdl, text):
if not tts_mdl or not text: return
if not tts_mdl or not text:
return
bin = b""
for chunk in tts_mdl.tts(text):
bin += chunk
@ -641,7 +650,8 @@ def ask(question, kb_ids, tenant_id):
idx = set([kbinfos["chunks"][int(i)]["doc_id"] for i in idx])
recall_docs = [
d for d in kbinfos["doc_aggs"] if d["doc_id"] in idx]
if not recall_docs: recall_docs = kbinfos["doc_aggs"]
if not recall_docs:
recall_docs = kbinfos["doc_aggs"]
kbinfos["doc_aggs"] = recall_docs
refs = deepcopy(kbinfos)
for c in refs["chunks"]: