Instructions to use nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("/iopsstor/scratch/cscs/nmuendler/hf-cache/hub/models--deepseek-ai--DeepSeek-R1-Distill-Qwen-7B/snapshots/916b56a44061fd5cd7d6a8fb632557ed4f724f60")
model = PeftModel.from_pretrained(base_model, "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0")

Transformers

How to use nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 with Transformers:

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

pipe = pipeline("text-generation", model="nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0

SGLang

How to use nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 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 "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0" \
    --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": "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0",
		"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 "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0" \
        --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": "nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 with Docker Model Runner:
```
docker model run hf.co/nmuendler/DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0
```

DeepSeek-R1-Distill-Qwen-7B-rust-early-stop-run1-kl0 / run_config.json

nmuendler

Add files using upload-large-folder tool

86ec600 verified about 6 hours ago

Raw

History Blame Contribute Delete

1.79 kB

	{
	"base_model": "/iopsstor/scratch/cscs/nmuendler/hf-cache/hub/models--deepseek-ai--DeepSeek-R1-Distill-Qwen-7B/snapshots/916b56a44061fd5cd7d6a8fb632557ed4f724f60",
	"train_file": "datasets/training_set_filtered_code_max.jsonl",
	"validation_file": "datasets/rust_dataset_true_passed_test750_max.jsonl",
	"task": "rust",
	"reasoning": false,
	"reasoning_format": "think",
	"target_mode": "standard",
	"rust_prompt_variant": "current",
	"learning_rate": 0.0002,
	"epochs": 5,
	"batch_size": 2,
	"gradient_accumulation_steps": 8,
	"effective_batch_size": 16,
	"max_length": 3000,
	"lora_r": 64,
	"lora_alpha": 128,
	"lora_dropout": 0.05,
	"target_modules": [
	"q_proj",
	"k_proj",
	"v_proj",
	"o_proj",
	"gate_proj",
	"up_proj",
	"down_proj"
	],
	"full_finetune": false,
	"eval_step_fraction": 0.01,
	"eval_steps": 4,
	"validation_size": 100,
	"stop_reasoning_below": 90.0,
	"reasoning_probe_tokens": 20,
	"reasoning_probe_num_prompts": 100,
	"kl_coefficient": 0.0,
	"kl_temperature": 1.0,
	"kl_mask_mode": "full",
	"mask_end_think_labels": false,
	"gradient_checkpointing": false,
	"seed": 42,
	"pre_model_seed": null,
	"raw_training_examples": 6761,
	"filtered_training_examples": 6761,
	"optimizer_steps_per_epoch": 423,
	"train_file_sha256": "6c6fc896b9e48de81cce63881f5bab2497ab4bf6e212aec4202a38d0a7766450",
	"transformers_version": "5.16.1",
	"peft_version": "0.20.0",
	"torch_version": "2.13.0+cu130",
	"resolved_training_arguments": {
	"optim": "adamw_torch_fused",
	"lr_scheduler_type": "linear",
	"weight_decay": 0.0,
	"adam_beta1": 0.9,
	"adam_beta2": 0.999,
	"adam_epsilon": 1e-08,
	"max_grad_norm": 1.0,
	"warmup_ratio": null,
	"warmup_steps": 0,
	"seed": 42,
	"data_seed": 42
	}
	}