Instructions to use Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096 with Transformers:

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

pipe = pipeline("text-generation", model="Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096")
model = AutoModelForCausalLM.from_pretrained("Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096", device_map="auto")
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 Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096

SGLang

How to use Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096 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 "Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096" \
    --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": "Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096",
		"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 "Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096" \
        --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": "Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096 with Docker Model Runner:
```
docker model run hf.co/Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096
```

SmolLM2-135M Reasoning Recovery 30K — 4096

Continued full-parameter supervised fine-tuning from Ma7ee7/SmolLM2-135M-Reasoning-5K.

Reasoning-token format

<think> and </think> are atomic non-special added tokens. This keeps the markers visible to normal decoders and reasoning-aware chat interfaces.

Expected assistant response:

<think>
Short, relevant reasoning.
</think>
Final answer

Training

Context length: 4096
Train rows: 28,786
Validation rows: 1,000
Learning rate: 1.5e-05
Epochs: 2.0
Reasoning loss weight: 1.0
Final-answer loss weight: 4.0
Seed: 3407

Downloads last month: 11

Safetensors

Model size

0.1B params

Tensor type

BF16

Model tree for Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096

Base model

HuggingFaceTB/SmolLM2-135M

Quantized

HuggingFaceTB/SmolLM2-135M-Instruct

Finetuned

Ma7ee7/SmolLM2-135M-Reasoning-5K

Finetuned

(1)

this model

Ma7ee7
/

SmolLM2-135M-Reasoning-Recovery-30K-4096

SmolLM2-135M Reasoning Recovery 30K — 4096

Reasoning-token format

Training

Model tree for Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096

Datasets used to train Ma7ee7/SmolLM2-135M-Reasoning-Recovery-30K-4096