Spaces:
Runtime error
Runtime error
changed to mistral model
Browse files- src/chatbot.py +8 -3
src/chatbot.py
CHANGED
@@ -12,11 +12,14 @@ import os
|
|
12 |
#load_dotenv(find_dotenv())
|
13 |
|
14 |
|
|
|
15 |
embeddings = HuggingFaceEmbeddings(model_name="paraphrase-multilingual-MiniLM-L12-v2")
|
16 |
llm = HuggingFaceHub(
|
17 |
# Try different model here
|
18 |
-
|
19 |
-
repo_id="CohereForAI/c4ai-command-r-v01",
|
|
|
|
|
20 |
task="text-generation",
|
21 |
model_kwargs={
|
22 |
"max_new_tokens": 512,
|
@@ -24,6 +27,8 @@ llm = HuggingFaceHub(
|
|
24 |
"temperature": 0.1,
|
25 |
"repetition_penalty": 1.03,
|
26 |
}
|
|
|
|
|
27 |
)
|
28 |
# To Do: Experiment with different templates replying in german or english depending on the input language
|
29 |
prompt1 = ChatPromptTemplate.from_template("""<s>[INST]
|
@@ -55,7 +60,7 @@ db = get_vectorstore(embeddings=embeddings, folder_path=folder_path, index_name=
|
|
55 |
|
56 |
def chatbot(message, history, db=db, llm=llm, prompt=prompt2):
|
57 |
raw_response = RAG(llm=llm, prompt=prompt, db=db, question=message)
|
58 |
-
response = raw_response['answer']
|
59 |
return response
|
60 |
|
61 |
# Retrieve speech contents based on keywords
|
|
|
12 |
#load_dotenv(find_dotenv())
|
13 |
|
14 |
|
15 |
+
|
16 |
embeddings = HuggingFaceEmbeddings(model_name="paraphrase-multilingual-MiniLM-L12-v2")
|
17 |
llm = HuggingFaceHub(
|
18 |
# Try different model here
|
19 |
+
repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
|
20 |
+
# repo_id="CohereForAI/c4ai-command-r-v01", # too large 69gb
|
21 |
+
# repo_id="CohereForAI/c4ai-command-r-v01-4bit", # too large 22 gb
|
22 |
+
# repo_id="meta-llama/Meta-Llama-3-8B", # too large 16 gb
|
23 |
task="text-generation",
|
24 |
model_kwargs={
|
25 |
"max_new_tokens": 512,
|
|
|
27 |
"temperature": 0.1,
|
28 |
"repetition_penalty": 1.03,
|
29 |
}
|
30 |
+
#,huggingfacehub_api_token
|
31 |
+
|
32 |
)
|
33 |
# To Do: Experiment with different templates replying in german or english depending on the input language
|
34 |
prompt1 = ChatPromptTemplate.from_template("""<s>[INST]
|
|
|
60 |
|
61 |
def chatbot(message, history, db=db, llm=llm, prompt=prompt2):
|
62 |
raw_response = RAG(llm=llm, prompt=prompt, db=db, question=message)
|
63 |
+
response = raw_response['answer'].split("Antwort: ")[1]
|
64 |
return response
|
65 |
|
66 |
# Retrieve speech contents based on keywords
|