Skip to content

Drop in a screenshot and convert it to clean code (HTML/Tailwind/React/Vue)

Python

add support for o1 but don't enable it (it's really slow and not everyone has access to it)

Abi Rajaadf50c90171e
diff --git a/backend/llm.py b/backend/llm.py
index 0f224a11..244f0944 100644
--- a/backend/llm.py
+++ b/backend/llm.py
@@ -50,35 +50,42 @@ async def stream_openai_response(
params = {
"model": model.value,
"messages": messages,
- "stream": True,
"timeout": 600,
- "temperature": 0.0,
}
- # Add 'max_tokens' only if the model is a GPT4 vision or Turbo model
- if (
- model == Llm.GPT_4_VISION
- or model == Llm.GPT_4_TURBO_2024_04_09
- or model == Llm.GPT_4O_2024_05_13
- ):
+ # O1 doesn't support streaming or temperature
+ if model != Llm.O1_2024_12_17:
+ params["temperature"] = 0
+ params["stream"] = True
+
+ # Add 'max_tokens' corresponding to the model
+ if model == Llm.GPT_4O_2024_05_13:
params["max_tokens"] = 4096
if model == Llm.GPT_4O_2024_11_20:
params["max_tokens"] = 16384
- stream = await client.chat.completions.create(**params) # type: ignore
- full_response = ""
- async for chunk in stream: # type: ignore
- assert isinstance(chunk, ChatCompletionChunk)
- if (
- chunk.choices
- and len(chunk.choices) > 0
- and chunk.choices[0].delta
- and chunk.choices[0].delta.content
- ):
- content = chunk.choices[0].delta.content or ""
- full_response += content
- await callback(content)
+ if model == Llm.O1_2024_12_17:
+ params["max_completion_tokens"] = 20000
+
+ # O1 doesn't support streaming
+ if model == Llm.O1_2024_12_17:
+ response = await client.chat.completions.create(**params) # type: ignore
+ full_response = response.choices[0].message.content # type: ignore
+ else:
+ stream = await client.chat.completions.create(**params) # type: ignore
+ full_response = ""
+ async for chunk in stream: # type: ignore
+ assert isinstance(chunk, ChatCompletionChunk)
+ if (
+ chunk.choices
+ and len(chunk.choices) > 0
+ and chunk.choices[0].delta
+ and chunk.choices[0].delta.content
+ ):
+ content = chunk.choices[0].delta.content or ""
+ full_response += content
+ await callback(content)
await client.close()
diff --git a/backend/routes/generate_code.py b/backend/routes/generate_code.py
index e1c9d98d..4afb413a 100644
--- a/backend/routes/generate_code.py
+++ b/backend/routes/generate_code.py
@@ -285,7 +285,7 @@ async def stream_code(websocket: WebSocket):
tasks: List[Coroutine[Any, Any, Completion]] = []
for index, model in enumerate(variant_models):
- if model == Llm.GPT_4O_2024_11_20:
+ if model == Llm.GPT_4O_2024_11_20 or model == Llm.O1_2024_12_17:
if openai_api_key is None:
await throw_error("OpenAI API key is missing.")
raise Exception("OpenAI API key is missing.")
@@ -296,7 +296,7 @@ async def stream_code(websocket: WebSocket):
api_key=openai_api_key,
base_url=openai_base_url,
callback=lambda x, i=index: process_chunk(x, i),
- model=Llm.GPT_4O_2024_11_20,
+ model=model,
)
)
elif model == Llm.GEMINI_2_0_FLASH_EXP and GEMINI_API_KEY: