Instructions to use nisten/tqwendo-36b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nisten/tqwendo-36b with Transformers:

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

pipe = pipeline("text-generation", model="nisten/tqwendo-36b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nisten/tqwendo-36b")
model = AutoModelForCausalLM.from_pretrained("nisten/tqwendo-36b")
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 Settings

vLLM

How to use nisten/tqwendo-36b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nisten/tqwendo-36b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nisten/tqwendo-36b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nisten/tqwendo-36b

SGLang

How to use nisten/tqwendo-36b 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 "nisten/tqwendo-36b" \
    --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": "nisten/tqwendo-36b",
		"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 "nisten/tqwendo-36b" \
        --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": "nisten/tqwendo-36b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nisten/tqwendo-36b with Docker Model Runner:
```
docker model run hf.co/nisten/tqwendo-36b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

This is a competitive coding model that should be better than qwen-coder-32b-instruct which we're running on github.gg

36B experimental merge was to fix the repetition issues with the huihui-ai coder abliterated.

There is a draft model to go with this one for speculative decoding and chain of thought reasoning: https://huggingface.co/nisten/qwen2.5-coder-7b-abliterated-128k-AWQ

Using the above 4bit 7b in conjuction with the 36b is meant to setup a chain-of-thought reasoner, evaluator similar to what O1-O3 is probably doing. This way the 7b 4bit only uses up an extra 4-6Gb on the gpu, but greatly both speeds up speculative decoding AND also chain-of-throught evals.

I.e. in this case the model was able to write an almost working heh.. chat interface for itself in one shot. And this was the WHOLE interface, including python and html and styling and api calls.