UniqAI/open-webui
1
0
Fork 0
mirror of https://github.com/open-webui/open-webui synced 2025-02-22 13:18:25 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

Added OpenAI usagerequested keys

Browse Source
This commit is contained in:
Simone 2025-02-19 15:28:39 +01:00
parent 8f72f8def9
commit 074ce356de
1 changed files with 26 additions and 51 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

77
backend/open_webui/utils/response.py
View File

@ -23,18 +23,8 @@ def convert_ollama_tool_call_to_openai(tool_calls: dict) -> dict:
openai_tool_calls.append(openai_tool_call) openai_tool_calls.append(openai_tool_call)
return openai_tool_calls return openai_tool_calls
def convert_ollama_usage_to_openai(data: dict) -> dict:
def convert_response_ollama_to_openai(ollama_response: dict) -> dict: return {
model = ollama_response.get("model", "ollama")
message_content = ollama_response.get("message", {}).get("content", "")
tool_calls = ollama_response.get("message", {}).get("tool_calls", None)
openai_tool_calls = None
if tool_calls:
openai_tool_calls = convert_ollama_tool_call_to_openai(tool_calls)
data = ollama_response
usage = {
"response_token/s": ( "response_token/s": (
round( round(
( (
@ -66,14 +56,37 @@ def convert_response_ollama_to_openai(ollama_response: dict) -> dict:
"total_duration": data.get("total_duration", 0), "total_duration": data.get("total_duration", 0),
"load_duration": data.get("load_duration", 0), "load_duration": data.get("load_duration", 0),
"prompt_eval_count": data.get("prompt_eval_count", 0), "prompt_eval_count": data.get("prompt_eval_count", 0),
"prompt_tokens": int(data.get("prompt_eval_count", 0)), # This is the OpenAI compatible key
"prompt_eval_duration": data.get("prompt_eval_duration", 0), "prompt_eval_duration": data.get("prompt_eval_duration", 0),
"eval_count": data.get("eval_count", 0), "eval_count": data.get("eval_count", 0),
"completion_tokens": int(data.get("eval_count", 0)), # This is the OpenAI compatible key
"eval_duration": data.get("eval_duration", 0), "eval_duration": data.get("eval_duration", 0),
"approximate_total": (lambda s: f"{s // 3600}h{(s % 3600) // 60}m{s % 60}s")( "approximate_total": (lambda s: f"{s // 3600}h{(s % 3600) // 60}m{s % 60}s")(
(data.get("total_duration", 0) or 0) // 1_000_000_000 (data.get("total_duration", 0) or 0) // 1_000_000_000
), ),
"total_tokens": int( # This is the OpenAI compatible key
data.get("prompt_eval_count", 0) + data.get("eval_count", 0)
),
"completion_tokens_details": { # This is the OpenAI compatible key
"reasoning_tokens": 0,
"accepted_prediction_tokens": 0,
"rejected_prediction_tokens": 0
}
} }
def convert_response_ollama_to_openai(ollama_response: dict) -> dict:
model = ollama_response.get("model", "ollama")
message_content = ollama_response.get("message", {}).get("content", "")
tool_calls = ollama_response.get("message", {}).get("tool_calls", None)
openai_tool_calls = None
if tool_calls:
openai_tool_calls = convert_ollama_tool_call_to_openai(tool_calls)
data = ollama_response
usage = convert_ollama_usage_to_openai(data)
response = openai_chat_completion_message_template( response = openai_chat_completion_message_template(
model, message_content, openai_tool_calls, usage model, message_content, openai_tool_calls, usage
) )
@ -96,45 +109,7 @@ async def convert_streaming_response_ollama_to_openai(ollama_streaming_response)
usage = None usage = None
if done: if done:
usage = { usage = convert_ollama_usage_to_openai(data)
"response_token/s": (
round(
(
(
data.get("eval_count", 0)
/ ((data.get("eval_duration", 0) / 10_000_000))
)
* 100
),
2,
)
if data.get("eval_duration", 0) > 0
else "N/A"
),
"prompt_token/s": (
round(
(
(
data.get("prompt_eval_count", 0)
/ ((data.get("prompt_eval_duration", 0) / 10_000_000))
)
* 100
),
2,
)
if data.get("prompt_eval_duration", 0) > 0
else "N/A"
),
"total_duration": data.get("total_duration", 0),
"load_duration": data.get("load_duration", 0),
"prompt_eval_count": data.get("prompt_eval_count", 0),
"prompt_eval_duration": data.get("prompt_eval_duration", 0),
"eval_count": data.get("eval_count", 0),
"eval_duration": data.get("eval_duration", 0),
"approximate_total": (
lambda s: f"{s // 3600}h{(s % 3600) // 60}m{s % 60}s"
)((data.get("total_duration", 0) or 0) // 1_000_000_000),
}
data = openai_chat_chunk_message_template( data = openai_chat_chunk_message_template(
model, message_content if not done else None, openai_tool_calls, usage model, message_content if not done else None, openai_tool_calls, usage