Phi-4

Run Microsoft's Phi-4 small language model on Clore.ai GPUs

Run Microsoft's Phi-4 - a small but powerful language model.

All examples can be run on GPU servers rented through CLORE.AI Marketplace.

Renting on CLORE.AI

1. Visit CLORE.AI Marketplace

2. Filter by GPU type, VRAM, and price

3. Choose On-Demand (fixed rate) or Spot (bid price)

4. Configure your order:

   • Select Docker image

   • Set ports (TCP for SSH, HTTP for web UIs)

   • Add environment variables if needed

   • Enter startup command

5. Select payment: CLORE, BTC, or USDT/USDC

6. Create order and wait for deployment

Access Your Server

• Find connection details in My Orders

• Web interfaces: Use the HTTP port URL

• SSH: ssh -p <port> root@<proxy-address>

What is Phi-4?

Phi-4 from Microsoft offers:

• 14B parameters with excellent performance

• Beats larger models on benchmarks

• Strong reasoning and math

• Efficient inference

Model Variants

Model	Parameters	VRAM	Specialty
Phi-4	14B	16GB	General
Phi-3.5-mini	3.8B	4GB	Lightweight
Phi-3.5-MoE	42B (6.6B active)	16GB	Mixture of Experts
Phi-3.5-vision	4.2B	6GB	Vision

Quick Deploy

Docker Image:

pytorch/pytorch:2.11.0-cuda12.8-cudnn9-runtime

Ports:

22/tcp
8000/http

Command:

pip install transformers accelerate torch && \
python phi4_server.py

Accessing Your Service

After deployment, find your http_pub URL in My Orders:

1. Go to My Orders page

2. Click on your order

3. Find the http_pub URL (e.g., abc123.clorecloud.net)

Use https://YOUR_HTTP_PUB_URL instead of localhost in examples below.

Using Ollama

# Run Phi-4
ollama run phi4

# Phi-3.5 mini (faster)
ollama run phi3.5

# Phi-3.5 vision
ollama run phi3.5-vision

Installation

pip install transformers accelerate torch

Basic Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "microsoft/Phi-4"

tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True
)

messages = [
    {"role": "system", "content": "You are a helpful AI assistant."},
    {"role": "user", "content": "Explain the difference between TCP and UDP."}
]

inputs = tokenizer.apply_chat_template(
    messages,
    return_tensors="pt",
    add_generation_prompt=True
).to("cuda")

outputs = model.generate(
    inputs,
    max_new_tokens=512,
    temperature=0.7,
    do_sample=True
)

response = tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)
print(response)

Phi-3.5-Vision

For image understanding:

from transformers import AutoModelForCausalLM, AutoProcessor
from PIL import Image
import torch

model_id = "microsoft/Phi-3.5-vision-instruct"

processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True
)

image = Image.open("diagram.png")

messages = [
    {"role": "user", "content": "<|image_1|>\nDescribe this diagram in detail."}
]

prompt = processor.tokenizer.apply_chat_template(
    messages, tokenize=False, add_generation_prompt=True
)

inputs = processor(prompt, [image], return_tensors="pt").to("cuda")

outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.7
)

response = processor.decode(outputs[0], skip_special_tokens=True)
print(response)

Math and Reasoning

messages = [
    {"role": "user", "content": """
Solve step by step:
A farmer has chickens and rabbits.
Total heads: 35
Total legs: 94
How many of each animal?
"""}
]

# Phi-4 excels at step-by-step reasoning

Code Generation

messages = [
    {"role": "user", "content": """
Write a Python implementation of binary search tree with:
- Insert
- Search
- Delete
- In-order traversal
Include type hints and docstrings.
"""}
]

Quantized Inference

from transformers import AutoModelForCausalLM, BitsAndBytesConfig
import torch

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16
)

model = AutoModelForCausalLM.from_pretrained(
    "microsoft/Phi-4",
    quantization_config=quantization_config,
    device_map="auto",
    trust_remote_code=True
)

Gradio Interface

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "microsoft/Phi-4"
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id, torch_dtype=torch.bfloat16, device_map="auto", trust_remote_code=True
)

def chat(message, history, system_prompt, temperature):
    messages = [{"role": "system", "content": system_prompt}]
    for h in history:
        messages.append({"role": "user", "content": h[0]})
        messages.append({"role": "assistant", "content": h[1]})
    messages.append({"role": "user", "content": message})

    inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to("cuda")
    outputs = model.generate(inputs, max_new_tokens=512, temperature=temperature, do_sample=True)

    return tokenizer.decode(outputs[0][inputs.shape[1]:], skip_special_tokens=True)

demo = gr.ChatInterface(
    fn=chat,
    additional_inputs=[
        gr.Textbox(value="You are a helpful assistant.", label="System"),
        gr.Slider(0.1, 1.5, value=0.7, label="Temperature")
    ],
    title="Phi-4 Chat"
)

demo.launch(server_name="0.0.0.0", server_port=7860)

Performance

Model	GPU	Tokens/sec
Phi-3.5-mini	RTX 3060	~100
Phi-3.5-mini	RTX 4090	~150
Phi-4	RTX 4090	~60
Phi-4	A100	~90
Phi-4 (4-bit)	RTX 3090	~40

Benchmarks

Model	MMLU	HumanEval	GSM8K
Phi-4	84.8%	82.6%	94.6%
GPT-4-Turbo	86.4%	85.4%	94.2%
Llama-3.1-70B	83.6%	80.5%	92.1%

Phi-4 matches or beats much larger models

Troubleshooting

"trust_remote_code" error

• Add trust_remote_code=True to from_pretrained()

• This is required for Phi models

Repetitive outputs

• Lower temperature (0.3-0.6)

• Add repetition_penalty=1.1

• Use proper chat template

Memory issues

• Phi-4 is efficient but still needs ~8GB for 14B

• Use 4-bit quantization if needed

• Reduce context length

Wrong output format

• Use apply_chat_template() for proper formatting

• Check you're using instruct version, not base

Cost Estimate

Typical CLORE.AI marketplace rates (as of 2024):

GPU	Hourly Rate	Daily Rate	4-Hour Session
RTX 3060	~$0.03	~$0.70	~$0.12
RTX 3090	~$0.06	~$1.50	~$0.25
RTX 4090	~$0.10	~$2.30	~$0.40
A100 40GB	~$0.17	~$4.00	~$0.70
A100 80GB	~$0.25	~$6.00	~$1.00

Prices vary by provider and demand. Check CLORE.AI Marketplace for current rates.

Save money:

• Use the Spot market for interruptible work — about a third of servers price spot below on-demand (median ~13% off), the rest match it

• Pay with CLORE tokens

• Compare prices across different providers

Use Cases

• Math tutoring

• Code assistance

• Document analysis (vision)

• Efficient edge deployment

• Cost-effective inference

Next Steps

• Qwen2.5 - Alternative model

• Gemma 2 - Google's model

• Llama 3.2 - Meta's model