Instructions to use KhushalM/Qwen2.5-1.5-SFT-Merged with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use KhushalM/Qwen2.5-1.5-SFT-Merged with Transformers:

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

pipe = pipeline("text-generation", model="KhushalM/Qwen2.5-1.5-SFT-Merged")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("KhushalM/Qwen2.5-1.5-SFT-Merged")
model = AutoModelForCausalLM.from_pretrained("KhushalM/Qwen2.5-1.5-SFT-Merged", 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 KhushalM/Qwen2.5-1.5-SFT-Merged with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "KhushalM/Qwen2.5-1.5-SFT-Merged"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "KhushalM/Qwen2.5-1.5-SFT-Merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/KhushalM/Qwen2.5-1.5-SFT-Merged

SGLang

How to use KhushalM/Qwen2.5-1.5-SFT-Merged 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 "KhushalM/Qwen2.5-1.5-SFT-Merged" \
    --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": "KhushalM/Qwen2.5-1.5-SFT-Merged",
		"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 "KhushalM/Qwen2.5-1.5-SFT-Merged" \
        --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": "KhushalM/Qwen2.5-1.5-SFT-Merged",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use KhushalM/Qwen2.5-1.5-SFT-Merged with Docker Model Runner:
```
docker model run hf.co/KhushalM/Qwen2.5-1.5-SFT-Merged
```

KhushalM/Qwen2.5-1.5-SFT-Merged

This is a causal language model fine‑tuned to generate conceptual explanations in the style of Richard Feynman, using supervised learning on curated prompt–completion pairs. It serves as a strong pedagogical baseline before any reinforcement learning stage.

Model Details

Model Description

“KhushalM/Qwen2.5-1.5-SFT-Merged” is built on top of the Qwen/Qwen2.5-1.5B-Instruct base. It was supervised‑finetuned on approximately 750 high‑quality prompt–completion pairs designed to replicate the Feynman teaching approach: simple, first‑principles explanations with concrete analogies, layered structure, and occasional comprehension checks. After finetuning, LoRA adapters were merged for streamlined inference.

Finetuned by: Khushal Mandavia
Model type: Decoder‑only transformer (causal LM)
Finetuned from: Qwen/Qwen2.5-1.5B-Instruct
Repository: https://huggingface.co/KhushalM/Qwen2.5-1.5-SFT-Merged

Uses

Direct Use

Conceptual Explanation: Generate analogy‑driven, step‑by‑step explanations of scientific and technical topics.
Educational Prototyping: Rapidly test teaching content formats or draft study materials.

Downstream Use

Warm‑start for RL: Serve as the supervised-learning base before reinforcement learning stages.
Fine‑tuned Deployment: Use in chatbots or tutoring systems requiring clear pedagogical style.

Out‑of‑Scope Use

High‑stakes Factual Tasks: Not designed for critical decision‑making without verification.
General Chat: Optimized for explanatory style rather than casual conversation.

Bias, Risks, and Limitations

The model produces engaging and clear explanations but may:

Overgeneralize: Simplifications can omit important nuances.
Hallucinate: Introduce plausible but incorrect statements.
Style Fixation: Rigidly adhere to the Feynman pattern even when another tone is preferable.

Recommendation: Always review outputs for accuracy and context relevance. Consider further domain‑specific finetuning where needed.

How to Get Started with the Model

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

model_id = "KhushalM/Qwen2.5-1.5-SFT-Merged"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", trust_remote_code=True)

generator = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    device="cuda"  # or device="cpu"
)

prompt = "Explain the concept of entropy from first principles."
output = generator(prompt, max_new_tokens=180, temperature=0.7, top_p=0.9)
print(output[0]["generated_text"])

Downloads last month: 8

Safetensors

Model size

2B params

Tensor type

F32

Model tree for KhushalM/Qwen2.5-1.5-SFT-Merged

Finetunes

2 models