Instructions to use microsoft/FastContext-1.0-4B-RL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use microsoft/FastContext-1.0-4B-RL with Transformers:

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

pipe = pipeline("text-generation", model="microsoft/FastContext-1.0-4B-RL")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("microsoft/FastContext-1.0-4B-RL")
model = AutoModelForMultimodalLM.from_pretrained("microsoft/FastContext-1.0-4B-RL")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use microsoft/FastContext-1.0-4B-RL with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "microsoft/FastContext-1.0-4B-RL"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "microsoft/FastContext-1.0-4B-RL",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/microsoft/FastContext-1.0-4B-RL

SGLang

How to use microsoft/FastContext-1.0-4B-RL 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 "microsoft/FastContext-1.0-4B-RL" \
    --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": "microsoft/FastContext-1.0-4B-RL",
		"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 "microsoft/FastContext-1.0-4B-RL" \
        --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": "microsoft/FastContext-1.0-4B-RL",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use microsoft/FastContext-1.0-4B-RL with Docker Model Runner:
```
docker model run hf.co/microsoft/FastContext-1.0-4B-RL
```

FastContext-1.0-4B-RL in production agent integration

by ghostwithahat - opened 3 days ago

Discussion

ghostwithahat

3 days ago

I tested FastContext-1.0-4B-RL as an explore_repository subagent in my Go coding agent (ahle). Served via llama.cpp on RTX 3090, unquantized, temp=0.0, 104K ctx. Compared head-to-head against FastContext-1.0-4B-SFT on the same codebase (~450 files, ~90K LOC Go).

RL improvements over SFT:

2–3× faster per loop (8–14s vs 19–26s)
Better citation precision overall
Main agent re-reads cited files ~60% less often
Fewer hallucinated file paths

Remaining issues (both variants):

<final_answer> XML tags are unreliable — the model often produces correct citation text but omits the tags. I had to add a regex fallback for lines matching /path/file.go:line-range.
Path hallucination reappears even at temp=0.0 (e.g. /home/ss/ahle/ vs the correct /home/ss/ai/ahle/), likely triggered by long directory listings in the system prompt.
Line ranges remain broad (file.go:1-280 when the answer is on lines 5–20).
Performance varies run-to-run: 11 tool calls in the best test, 13 in the worst. SFT averaged 29.

Bottom line: FastContext-1.0-4B-RL is a clear step up from SFT and usable with a fallback parser. The tag consistency and citation precision would benefit from more RL iterations. Thanks for releasing both variants — looking forward to future iterations.

I'll keep the FastContext subagant active in my setup and will report back if I gather more systematic data over time.

Thank's a lot and best wishes from lower franconia.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment