Instructions to use AlexL0701/broken-model-fixed with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AlexL0701/broken-model-fixed with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="AlexL0701/broken-model-fixed")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("AlexL0701/broken-model-fixed")
model = AutoModelForCausalLM.from_pretrained("AlexL0701/broken-model-fixed")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use AlexL0701/broken-model-fixed with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AlexL0701/broken-model-fixed"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AlexL0701/broken-model-fixed",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AlexL0701/broken-model-fixed

SGLang

How to use AlexL0701/broken-model-fixed with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "AlexL0701/broken-model-fixed" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AlexL0701/broken-model-fixed",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "AlexL0701/broken-model-fixed" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AlexL0701/broken-model-fixed",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use AlexL0701/broken-model-fixed with Docker Model Runner:
```
docker model run hf.co/AlexL0701/broken-model-fixed
```

This repository fixes the original broken-model so it works correctly with OpenAI-style chat APIs such as /chat/completions.

The problem was that the original tokenizer configuration did not define a chat template. When a model is served with Text Generation Inference, chat requests are sent as structured messages with roles like user and assistant. Without a chat template, the server does not know how to convert those messages into the text format expected by the model, which causes chat requests to fail.

To fix this, tokenizer_config.json was updated. The bos_token was set to <|endoftext|> instead of null, and a chat_template was added. The chat template formats each message using <|im_start|> and <|im_end|> tokens and includes the role and content of each message. This matches the expected Qwen-style prompt format.

With these changes, the server can correctly serialize chat messages into a prompt and the model can generate responses normally when using /chat/completions.

Downloads last month: 4

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for AlexL0701/broken-model-fixed

Base model

meta-llama/Llama-3.1-8B

Finetuned

(1787)

this model

AlexL0701
/

broken-model-fixed

Model tree for AlexL0701/broken-model-fixed

Space using AlexL0701/broken-model-fixed 1