Skip to content

feat: add automatic language detection for responses #144

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Libres-coder wants to merge 2 commits into
base: main
Choose a base branch
from
Open

feat: add automatic language detection for responses #144

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a037ee9
feat: add automatic language detection for responses
Libres-coder Nov 12, 2025
4ee6b38
add
Libres-coder Nov 12, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions deepseek_vl2/serve/inference.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,6 +47,12 @@ def load_model(model_path, dtype=torch.bfloat16):
def convert_conversation_to_prompts(conversation: Conversation):
conv_prompts = []

if conversation.system_message and conversation.system_message.strip():
conv_prompts.append({
"role": "system",
"content": conversation.system_message
})

last_image = None

messages = conversation.messages
Expand Down
15 changes: 13 additions & 2 deletions web_demo.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@
import sys
import base64
from PIL import Image
from langdetect import detect

import gradio as gr
import torch
Expand Down Expand Up @@ -148,7 +149,7 @@ def fetch_model(model_name: str, dtype=torch.bfloat16):


def generate_prompt_with_history(
text, images, history, vl_chat_processor, tokenizer, max_length=2048
text, images, history, vl_chat_processor, tokenizer, max_length=2048, language="en"
):
"""
Generate a prompt with history for the deepseek application.
Expand All @@ -159,6 +160,7 @@ def generate_prompt_with_history(
history (list): List of previous conversation messages.
tokenizer: The tokenizer used for encoding the prompt.
max_length (int): The maximum length of the prompt.
language (str): The detected language code for setting response language.

Returns:
tuple: A tuple containing the generated prompt, image list, conversation, and conversation copy. If the prompt could not be generated within the max_length limit, returns None.
Expand All @@ -171,6 +173,7 @@ def generate_prompt_with_history(

# Initialize conversation
conversation = vl_chat_processor.new_chat_template()
conversation.set_system_message(f"Please reply in {language} language.")

if history:
conversation.messages = history
Expand Down Expand Up @@ -342,6 +345,14 @@ def predict(
if images is None:
images = []

detected_language = "en"
try:
detected_language = detect(text)
if detected_language in ("zh-cn", "zh-tw"):
detected_language = "zh"
except Exception as e:
logger.warning(f"Language detection failed: {e}. Defaulting to English.")

# load images
pil_images = []
for img_or_file in images:
Expand All @@ -362,6 +373,7 @@ def predict(
vl_chat_processor,
tokenizer,
max_length=max_context_length_tokens,
language=detected_language
)
all_conv, last_image = convert_conversation_to_prompts(conversation)

Expand Down Expand Up @@ -449,7 +461,6 @@ def retry(
max_length_tokens,
max_context_length_tokens,
model_select_dropdown,
args.chunk_size
)


Expand Down