Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -482,7 +482,31 @@ This information is from {date}.
|
|
482 |
You are an expert AI assistant. Write a detailed summary of the information provided in this source that is relevant to the following user request: '{query}'
|
483 |
Base your summary strictly on the information from this source. Only include information that is directly supported by the given content.
|
484 |
If any part of the information cannot be verified from this source, clearly state that it could not be confirmed."""
|
485 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
486 |
# Generate an overall summary after processing all sources
|
487 |
overall_prompt = f"""Based on the summaries you've generated for each source: '{accumulated_response}', provide a concise overall summary that addresses the user's query: '{query}'
|
488 |
Highlight any conflicting information or gaps in the available data."""
|
|
|
482 |
You are an expert AI assistant. Write a detailed summary of the information provided in this source that is relevant to the following user request: '{query}'
|
483 |
Base your summary strictly on the information from this source. Only include information that is directly supported by the given content.
|
484 |
If any part of the information cannot be verified from this source, clearly state that it could not be confirmed."""
|
485 |
+
|
486 |
+
if model == "@cf/meta/llama-3.1-8b-instruct":
|
487 |
+
# Use Cloudflare API
|
488 |
+
source_response = ""
|
489 |
+
for response in get_response_from_cloudflare(prompt="", context=context, query=query, num_calls=1, temperature=temperature, search_type="web"):
|
490 |
+
source_response += response
|
491 |
+
accumulated_response += f"Source {i} ({source}):\n\n{source_response}\n\n"
|
492 |
+
yield accumulated_response, ""
|
493 |
+
else:
|
494 |
+
# Use Hugging Face API
|
495 |
+
client = InferenceClient(model, token=huggingface_token)
|
496 |
+
|
497 |
+
source_response = ""
|
498 |
+
for message in client.chat_completion(
|
499 |
+
messages=[{"role": "user", "content": prompt}],
|
500 |
+
max_tokens=2000,
|
501 |
+
temperature=temperature,
|
502 |
+
stream=True,
|
503 |
+
):
|
504 |
+
if message.choices and message.choices[0].delta and message.choices[0].delta.content:
|
505 |
+
chunk = message.choices[0].delta.content
|
506 |
+
source_response += chunk
|
507 |
+
accumulated_response += f"Source {i} ({source}):\n\n{source_response}\n\n"
|
508 |
+
yield accumulated_response, ""
|
509 |
+
|
510 |
# Generate an overall summary after processing all sources
|
511 |
overall_prompt = f"""Based on the summaries you've generated for each source: '{accumulated_response}', provide a concise overall summary that addresses the user's query: '{query}'
|
512 |
Highlight any conflicting information or gaps in the available data."""
|