Instructions to use MaziyarPanahi/calme-2.2-phi3-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MaziyarPanahi/calme-2.2-phi3-4b with Transformers:

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

pipe = pipeline("text-generation", model="MaziyarPanahi/calme-2.2-phi3-4b", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MaziyarPanahi/calme-2.2-phi3-4b", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("MaziyarPanahi/calme-2.2-phi3-4b", trust_remote_code=True)
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

vLLM

How to use MaziyarPanahi/calme-2.2-phi3-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MaziyarPanahi/calme-2.2-phi3-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MaziyarPanahi/calme-2.2-phi3-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MaziyarPanahi/calme-2.2-phi3-4b

SGLang

How to use MaziyarPanahi/calme-2.2-phi3-4b 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 "MaziyarPanahi/calme-2.2-phi3-4b" \
    --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": "MaziyarPanahi/calme-2.2-phi3-4b",
		"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 "MaziyarPanahi/calme-2.2-phi3-4b" \
        --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": "MaziyarPanahi/calme-2.2-phi3-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use MaziyarPanahi/calme-2.2-phi3-4b with Docker Model Runner:
```
docker model run hf.co/MaziyarPanahi/calme-2.2-phi3-4b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

MaziyarPanahi/calme-2.2-phi3-4b

This model is a fine-tune (DPO) of microsoft/Phi-3-mini-4k-instruct model.

⚡ Quantized GGUF

All GGUF models are available here: MaziyarPanahi/calme-2.2-phi3-4b-GGUF

🏆 Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	23.21
IFEval (0-Shot)	50.69
BBH (3-Shot)	37.73
MATH Lvl 5 (4-Shot)	2.34
GPQA (0-shot)	9.51
MuSR (0-shot)	7.70
MMLU-PRO (5-shot)	31.27

Metric	Value
Avg.	69.78
AI2 Reasoning Challenge (25-Shot)	62.80
HellaSwag (10-Shot)	80.76
MMLU (5-Shot)	69.10
TruthfulQA (0-shot)	59.97
Winogrande (5-shot)	72.45
GSM8k (5-shot)	73.62

Prompt Template

This model uses ChatML prompt template:

<|im_start|>system
{System}
<|im_end|>
<|im_start|>user
{User}
<|im_end|>
<|im_start|>assistant
{Assistant}

How to use

You can use this model by using MaziyarPanahi/calme-2.2-phi3-4b as the model name in Hugging Face's transformers library.

from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
from transformers import pipeline
import torch

model_id = "MaziyarPanahi/calme-2.2-phi3-4b"

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
    # attn_implementation="flash_attention_2"
)

tokenizer = AutoTokenizer.from_pretrained(
    model_id,
    trust_remote_code=True
)

streamer = TextStreamer(tokenizer)

messages = [
    {"role": "system", "content": "You are a pirate chatbot who always responds in pirate speak!"},
    {"role": "user", "content": "Who are you?"},
]

# this should work perfectly for the model to stop generating
terminators = [
    tokenizer.eos_token_id, # this should be <|im_end|>
    tokenizer.convert_tokens_to_ids("<|assistant|>"), # sometimes model stops generating at <|assistant|>
    tokenizer.convert_tokens_to_ids("<|end|>") # sometimes model stops generating at <|end|>
]

pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
)

generation_args = {
    "max_new_tokens": 500,
    "return_full_text": False,
    "temperature": 0.0,
    "do_sample": False,
    "streamer": streamer,
    "eos_token_id": terminators,
}

output = pipe(messages, **generation_args)
print(output[0]['generated_text'])

Downloads last month: 541

Safetensors

Model size

4B params

Tensor type

BF16

Model tree for MaziyarPanahi/calme-2.2-phi3-4b

Base model

microsoft/Phi-3-mini-4k-instruct

Finetuned

(847)

this model

Quantizations

3 models

Collections including MaziyarPanahi/calme-2.2-phi3-4b

Evaluation results

normalized accuracy on AI2 Reasoning Challenge (25-Shot)
test set Open LLM Leaderboard

62.800
normalized accuracy on HellaSwag (10-Shot)
validation set Open LLM Leaderboard

80.760
accuracy on MMLU (5-Shot)
test set Open LLM Leaderboard

69.100
mc2 on TruthfulQA (0-shot)
validation set Open LLM Leaderboard

59.970
accuracy on Winogrande (5-shot)
validation set Open LLM Leaderboard

72.450
accuracy on GSM8k (5-shot)
test set Open LLM Leaderboard

73.620
strict accuracy on IFEval (0-Shot)
Open LLM Leaderboard

50.690
normalized accuracy on BBH (3-Shot)
Open LLM Leaderboard

37.730
exact match on MATH Lvl 5 (4-Shot)
Open LLM Leaderboard

2.340
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

9.510
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

7.700
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

31.270