Hugging Face Guide 2025: Models, Spaces & Inference API

# Install

pip install transformers torch accelerate

# Run a text generation model

from transformers import pipeline

# Downloads ~280MB to ~/.cache/huggingface/hub/
generator = pipeline("text-generation", model="microsoft/Phi-3-mini-4k-instruct")
result = generator("The key advantage of transformers is", max_new_tokens=100)
print(result[0]["generated_text"])

# Lower-level: from_pretrained for full control

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "meta-llama/Llama-3.2-1B-Instruct"

tokenizer = AutoTokenizer.from_pretrained(model_id, token="hf_YOUR_TOKEN")
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    token="hf_YOUR_TOKEN",
    torch_dtype=torch.bfloat16,
    device_map="auto"  # uses GPU if available
)

inputs = tokenizer("Explain neural networks in simple terms:", return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

# Python with huggingface_hub

pip install huggingface_hub

from huggingface_hub import InferenceClient

client = InferenceClient(token="hf_YOUR_TOKEN")

# Text generation (serverless — shared GPU, may queue)
result = client.text_generation(
    "Translate to French: The weather is nice today.",
    model="mistralai/Mistral-7B-Instruct-v0.3",
    max_new_tokens=100
)
print(result)

# Dedicated Endpoints — always-on, your own GPU

# Create at: ui.endpoints.huggingface.co
# Then use the endpoint URL:

client = InferenceClient(
    model="https://YOUR-ENDPOINT.endpoints.huggingface.cloud"
)
result = client.text_generation("Hello, world!", max_new_tokens=50)

# Example: Gradio Space (app.py)

import gradio as gr
from transformers import pipeline

classifier = pipeline("sentiment-analysis")

def analyze(text):
    result = classifier(text)[0]
    return f"{result['label']} ({result['score']:.2f})"

demo = gr.Interface(fn=analyze, inputs="text", outputs="text")
demo.launch()

# Search: filter by tag "GGUF" on huggingface.co/models
# Popular providers: bartowski, TheBloke, ggml-org

# Download a single file with huggingface_hub CLI:
pip install "huggingface_hub[cli]"

huggingface-cli download \
  bartowski/Llama-3.2-3B-Instruct-GGUF \
  Llama-3.2-3B-Instruct-Q4_K_M.gguf \
  --local-dir ./models

# Then run with Ollama (ollama run) or llama.cpp:
# ./llama-cli -m models/Llama-3.2-3B-Instruct-Q4_K_M.gguf -p "Hello" -n 100