Neuronx model for togethercomputer/LLaMA-2-7B-32K

This repository contains are AWS Inferentia2 and neuronx compatible checkpoint for togethercomputer/LLaMA-2-7B-32K. You can find detailed information about the base model on its Model Card.

Usage with optimum-neuron


from optimum.neuron import pipeline

# Load pipeline from Hugging Face repository
pipe = pipeline("text-generation", "WaelDataReply/Llama-2-7b-32k-seqlen-4096-bs-2")

# We use the tokenizer's chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
messages = [
    {"role": "user", "content": "What is the difference between AWS Step Functions and Apache Airflow?"},
]
prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
# Run generation
outputs = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])

Compilation Arguments

compilation arguments

{
  "num_cores": 2,
  "auto_cast_type": "fp16"
}

input_shapes

{
  "sequence_length": 4096,
  "batch_size": 2
}