Using LLMs with Hugging Face

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline

# 1. Basic text generation with a small model
model_name = "distilgpt2"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

# Generate text
prompt = "The future of AI is"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(
    inputs.input_ids,
    max_length=50,
    num_return_sequences=1,
    temperature=0.7,
    top_p=0.9,
    do_sample=True
)

generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(generated_text)

# 2. Using a pipeline for simpler interface
generator = pipeline('text-generation', model=model_name)
result = generator(prompt, max_length=50, num_return_sequences=1)
print(result[0]['generated_text'])

# 3. Using a different task - sentiment analysis
sentiment_analyzer = pipeline('sentiment-analysis')
result = sentiment_analyzer("I really enjoyed this movie!")
print(result)  # [{'label': 'POSITIVE', 'score': 0.9998}]

# 4. For more powerful models (with appropriate hardware)
# Uncomment and use as needed:
'''
# For models requiring more memory, use 8-bit quantization
from transformers import BitsAndBytesConfig

quantization_config = BitsAndBytesConfig(
    load_in_8bit=True,
    llm_int8_threshold=6.0
)

# Load a larger model in 8-bit precision
model_name = "meta-llama/Llama-2-7b-chat-hf"  # Requires access permissions
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    device_map="auto",
    quantization_config=quantization_config
)

# Generate with the larger model
prompt = "Write a short poem about coding"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(
    inputs.input_ids,
    max_length=200,
    temperature=0.7,
    do_sample=True
)

generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(generated_text)
'''

pip install transformers torch
# For advanced features:
pip install accelerate bitsandbytes