README.md

---
library_name: transformers
tags:
- torchao
- phi
- phi4
- nlp
- code
- math
- chat
- conversational
license: mit
language:
- multilingual
base_model:
- microsoft/Phi-4-mini-instruct
pipeline_tag: text-generation
---

[Phi4-mini](https://huggingface.co/microsoft/Phi-4-mini-instruct) model quantized with [torchao](https://huggingface.co/docs/transformers/main/en/quantization/torchao) float8 dynamic activation and float8 weight quantization (per row granularity), by PyTorch team. Use it directly, or serve using [vLLM](https://docs.vllm.ai/en/latest/) with 36% VRAM reduction, 15-20% speedup and little to no accuracy impact on H100.


# Quantization Recipe

Install the required packages:

```
pip install git+https://github.com/huggingface/transformers@main
pip install --pre torchao --index-url https://download.pytorch.org/whl/nightly/cu126
pip install torch
pip install accelerate
```

Use the following code to get the quantized model:

```
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig

model_id = "microsoft/Phi-4-mini-instruct"

from torchao.quantization import Float8DynamicActivationFloat8WeightConfig, PerRow
quant_config = Float8DynamicActivationFloat8WeightConfig(granularity=PerRow())
quantization_config = TorchAoConfig(quant_type=quant_config)
quantized_model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype=torch.bfloat16, quantization_config=quantization_config)
tokenizer = AutoTokenizer.from_pretrained(model_id)

# Push to hub
USER_ID = "YOUR_USER_ID"
MODEL_NAME = model_id.split("/")[-1]
save_to = f"{USER_ID}/{MODEL_NAME}-float8dq"
quantized_model.push_to_hub(save_to, safe_serialization=False)
tokenizer.push_to_hub(save_to)

# Manual Testing
prompt = "Hey, are you conscious? Can you talk to me?"
messages = [
    {
        "role": "system",
        "content": "",
    },
    {"role": "user", "content": prompt},
]
templated_prompt = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
print("Prompt:", prompt)
print("Templated prompt:", templated_prompt)
inputs = tokenizer(
    templated_prompt,
    return_tensors="pt",
).to("cuda")
generated_ids = quantized_model.generate(**inputs, max_new_tokens=128)
output_text = tokenizer.batch_decode(
    generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print("Response:", output_text[0][len(prompt):])
```

# Serving with vllm

Need to install vllm nightly to get some recent changes:
```
pip install vllm --pre --extra-index-url https://wheels.vllm.ai/nightly
```

Then we can serve with the following command:
```
vllm serve pytorch/Phi-4-mini-instruct-float8dq --tokenizer microsoft/Phi-4-mini-instruct -O3
```

# Model Quality
We rely on [lm-evaluation-harness](https://github.com/EleutherAI/lm-evaluation-harness) to evaluate the quality of the quantized model.
Need to install lm-eval from source:
https://github.com/EleutherAI/lm-evaluation-harness#install


## baseline
```
lm_eval --model hf --model_args pretrained=microsoft/Phi-4-mini-instruct --tasks hellaswag --device cuda:0 --batch_size 8
```

## float8 dynamic activation and float8 weight quantization (float8dq)
```
lm_eval --model hf --model_args pretrained=pytorch/Phi-4-mini-instruct-float8dq --tasks hellaswag --device cuda:0 --batch_size 8
```

| Benchmark                        |                |                     |
|----------------------------------|----------------|---------------------|
|                                  | Phi-4 mini-Ins | phi4-mini-float8dq  | 
| **Popular aggregated benchmark** |                |                     |
| mmlu (0-shot)                    | 66.73          |  Pending            |
| mmlu_pro (5-shot)                | 46.43          |  Pending            |
| **Reasoning**                    |                |                     |
| arc_challenge (0-shot)           | 56.91          |  56.66              |
| gpqa_main_zeroshot               | 30.13          |  29.46              |
| HellaSwag                        | 54.57          |  54.55              |
| openbookqa                       | 33.00          |  33.60              |
| piqa (0-shot)	                   | 77.64          |  77.48              |
| social_iqa                       | 49.59          |  49.28              |
| truthfulqa_mc2 (0-shot)          | 48.39          |  48.09              |
| winogrande  (0-shot)             | 71.11          |  72.77              |
| **Multilingual**                 |                |                     |
| mgsm_en_cot_en                   | 60.8           |  60.0               |
| **Math**                         |                |                     |
| gsm8k (5-shot)                   | 81.88          |  80.89              |
| mathqa (0-shot)                  | 42.31          |  42.51              |
| **Overall**                      | **TODO**       | **TODO**            |

# Peak Memory Usage


## Results

| Benchmark        |                |                                |
|------------------|----------------|--------------------------------|
|                  | Phi-4 mini-Ins | Phi-4-mini-instruct-float8dq   | 
| Peak Memory (GB) | 8.91           | 5.70 (36% reduction)           |


## Benchmark Peak Memory

We can use the following code to get a sense of peak memory usage during inference:


```
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig

# use "microsoft/Phi-4-mini-instruct" or "pytorch/Phi-4-mini-instruct-float8dq"
model_id = "microsoft/Phi-4-mini-instruct"
quantized_model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype=torch.bfloat16)
tokenizer = AutoTokenizer.from_pretrained(model_id)

torch.cuda.reset_peak_memory_stats()

prompt = "Hey, are you conscious? Can you talk to me?"
messages = [
    {
        "role": "system",
        "content": "",
    },
    {"role": "user", "content": prompt},
]
templated_prompt = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
print("Prompt:", prompt)
print("Templated prompt:", templated_prompt)
inputs = tokenizer(
    templated_prompt,
    return_tensors="pt",
).to("cuda")
generated_ids = quantized_model.generate(**inputs, max_new_tokens=128)
output_text = tokenizer.batch_decode(
    generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print("Response:", output_text[0][len(prompt):])

mem = torch.cuda.max_memory_reserved() / 1e9
print(f"Peak Memory Usage: {mem:.02f} GB")
```

# Model Performance

## Results (H100 machine)
| Benchmark                        |                |                          |
|----------------------------------|----------------|--------------------------|
|                                  | Phi-4 mini-Ins | phi4-mini-float8dq       | 
| latency (batch_size=1)           | 1.64s          | 1.41s (16% speedup)      |
| latency (batch_size=128)         | 3.1s           | 2.72s (14% speedup)      |
| serving (num_prompts=1)          | 1.35 req/s     | 1.57 req/s (16% speedup) |
| serving (num_prompts=1000)       | 66.68 req/s    | 80.53 req/s (21% speedup)|

Note the result of latency (benchmark_latency) is in seconds, and serving (benchmark_serving) is in number of requests per second.

## benchmark_latency

Need to install vllm nightly to get some recent changes
```
pip install vllm --pre --extra-index-url https://wheels.vllm.ai/nightly
```

Get vllm source code:
```
git clone git@github.com:vllm-project/vllm.git
```

Run the following under `vllm` root folder:

### baseline
```
python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model microsoft/Phi-4-mini-instruct --batch-size 1
```

### float8dq
```
python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model pytorch/Phi-4-mini-instruct-float8dq --batch-size 1
```

## benchmark_serving

We also benchmarked the throughput in a serving environment.

Download sharegpt dataset: `wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json`
Other datasets can be found in: https://github.com/vllm-project/vllm/tree/main/benchmarks

Get vllm source code:
```
git clone git@github.com:vllm-project/vllm.git
```

Run the following under `vllm` root folder:

### baseline
Server:
```
vllm serve microsoft/Phi-4-mini-instruct --tokenizer microsoft/Phi-4-mini-instruct -O3
```

Client:
```
python benchmarks/benchmark_serving.py --backend vllm --dataset-name sharegpt --tokenizer microsoft/Phi-4-mini-instruct --dataset-path ./ShareGPT_V3_unfiltered_cleaned_split.json --model microsoft/Phi-4-mini-instruct --num-prompts 1
```

### float8dq
Server:
```
vllm serve pytorch/Phi-4-mini-instruct-float8dq --tokenizer microsoft/Phi-4-mini-instruct -O3
```

Client:
```
python benchmarks/benchmark_serving.py --backend vllm --dataset-name sharegpt --tokenizer microsoft/Phi-4-mini-instruct --dataset-path ./ShareGPT_V3_unfiltered_cleaned_split.json --model jerryzh168/phi4-mini-float8dq --num-prompts 1
```

# Disclaimer
PyTorch has not performed safety evaluations or red teamed the quantized models. Performance characteristics, outputs, and behaviors may differ from the original models. Users are solely responsible for selecting appropriate use cases, evaluating and mitigating for accuracy, safety, and fairness, ensuring security, and complying with all applicable laws and regulations.

Nothing contained in this Model Card should be interpreted as or deemed a restriction or modification to the licenses the models are released under, including any limitations of liability or disclaimers of warranties provided therein.