Skip to content

Agora-X/Atom

Repository files navigation

Multi-Modality

Atom

a suite of finetuned LLMs for atomically precise function calling 🧪

✅ Massive function calling dataset of over 20M samples.

✅ First Model: Atom-Z-Tiny - Zephr trained on 100k samples

✅ Vision function calling coming soon

Install

You can install the package using pip

pip install atom-torch

Usage

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

tokenizer = AutoTokenizer.from_pretrained("kye/Atom-Z-Tiny-7B")

model = AutoModelForCausalLM.from_pretrained(
  "kye/Atom-Z-Tiny-7B", 
  trust_remote_code=True, 
).to(device)

task = """


[INST] <<SYS>>
<function>Available functions:
<function>{
    "name": "generate_password",
    "description": "Generate a random password with specified criteria",
    "parameters": {
        "type": "object",
        "properties": {
            "length": {
                "type": "integer",
                "description": "The length of the password"
            },
            "include_numbers": {
                "type": "boolean",
                "description": "Include numbers in the password"
            },
            "include_special_characters": {
                "type": "boolean",
                "description": "Include special characters in the password"
            }
        },
        "required": [
            "length"
        ]
    }
}
<</SYS>>

I need a new password. Can you generate one for me? [/INST]


"""

input_ids = tokenizer.encode(task, return_tensors="pt")
output = model.generate(input_ids.to(device), max_length=128, temperature=0.7).cpu()
output_text = tokenizer.decode(output[0], skip_special_tokens=True)
print(output_text)

Training

  • To train the models, run accelerate config and enable DeepSpeed acceleration. deepspeed/zero3.json was the configuration file used for training.0
  • Then run accelerate launch accelerate launch finetune.py

We're finetuning this model on this dataset

sh./run.sh

Citation

@misc{peng2023yarn,
    title={YaRN: Efficient Context Window Extension of Large Language Models}, 
    author={Bowen Peng and Jeffrey Quesnelle and Honglu Fan and Enrico Shippole},
    year={2023},
    eprint={2309.00071},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}