You can deploy this model and start earning money today!

RichardErkhov/
Columbia-NLP_-_LION-LLaMA-3-8b-dpo-v1.0-gguf

Name: Columbia-NLP_-_LION-LLaMA-3-8b-dpo-v1.0-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

LION-LLaMA-3-8b-dpo-v1.0 - GGUF

Model creator: https://huggingface.co/Columbia-NLP/
Original model: https://huggingface.co/Columbia-NLP/LION-LLaMA-3-8b-dpo-v1.0/

Name	Quant method	Size
LION-LLaMA-3-8b-dpo-v1.0.Q2_K.gguf	Q2_K	2.96GB
LION-LLaMA-3-8b-dpo-v1.0.IQ3_XS.gguf	IQ3_XS	3.28GB
LION-LLaMA-3-8b-dpo-v1.0.IQ3_S.gguf	IQ3_S	3.43GB
LION-LLaMA-3-8b-dpo-v1.0.Q3_K_S.gguf	Q3_K_S	3.41GB
LION-LLaMA-3-8b-dpo-v1.0.IQ3_M.gguf	IQ3_M	3.52GB
LION-LLaMA-3-8b-dpo-v1.0.Q3_K.gguf	Q3_K	3.74GB
LION-LLaMA-3-8b-dpo-v1.0.Q3_K_M.gguf	Q3_K_M	3.74GB
LION-LLaMA-3-8b-dpo-v1.0.Q3_K_L.gguf	Q3_K_L	4.03GB
LION-LLaMA-3-8b-dpo-v1.0.IQ4_XS.gguf	IQ4_XS	4.18GB
LION-LLaMA-3-8b-dpo-v1.0.Q4_0.gguf	Q4_0	4.34GB
LION-LLaMA-3-8b-dpo-v1.0.IQ4_NL.gguf	IQ4_NL	4.38GB
LION-LLaMA-3-8b-dpo-v1.0.Q4_K_S.gguf	Q4_K_S	4.37GB
LION-LLaMA-3-8b-dpo-v1.0.Q4_K.gguf	Q4_K	4.58GB
LION-LLaMA-3-8b-dpo-v1.0.Q4_K_M.gguf	Q4_K_M	4.58GB
LION-LLaMA-3-8b-dpo-v1.0.Q4_1.gguf	Q4_1	4.78GB
LION-LLaMA-3-8b-dpo-v1.0.Q5_0.gguf	Q5_0	5.21GB
LION-LLaMA-3-8b-dpo-v1.0.Q5_K_S.gguf	Q5_K_S	5.21GB
LION-LLaMA-3-8b-dpo-v1.0.Q5_K.gguf	Q5_K	5.34GB
LION-LLaMA-3-8b-dpo-v1.0.Q5_K_M.gguf	Q5_K_M	5.34GB
LION-LLaMA-3-8b-dpo-v1.0.Q5_1.gguf	Q5_1	5.65GB
LION-LLaMA-3-8b-dpo-v1.0.Q6_K.gguf	Q6_K	6.14GB
LION-LLaMA-3-8b-dpo-v1.0.Q8_0.gguf	Q8_0	7.95GB

Original model description:

library_name: transformers tags: []

Model Card for LION-LLaMA-3-8b-dpo-v1.0

The LION-series are trained using an empirically optimized pipeline that consists of three stages: SFT, DPO, and online preference learning (online DPO). We find simple techniques such as sequence packing, loss masking in SFT, increasing the preference dataset size in DPO, and online DPO training can significantly improve the performance of language models. Our best models (the LION-series) exceed the performance of the official instruct models tuned with closed-source data and algorithms.

For training datasets, code, and evaluation scripts, please refer to our paper and codebase.

Model description

This model is finetuned from Columbia-NLP/LION-LLaMA-3-8b-sft-v1.0 using DPO from the LION pipeline.

Model type: meta-llama/Meta-Llama-3-8B
Language(s) (NLP): Primarily English
License: LLaMa-3 Terms of Use
Finetuned from model: Columbia-NLP/LION-LLaMA-3-8b-sft-v1.0

Performance

Model	Method	Size	Arena-Hard	AlpacaEval-2	MT-Bench	OpenLLM
LLaMA-3-8b	-	8B	-	-	-	63.05
LLaMA-3-8b-it	SFT+RS+DPO+PPO	8B	20.6	22.9	8.00	68.28
LION-LLaMA-3-8b-sft-v1.0 (ours)	SFT	8B	11.3	17.9	7.58	68.71
⮕ LION-LLaMA-3-8b-dpo-v1.0 (ours)	SFT+DPO	8B	19.1	21.8	8.12	71.28
LION-LLaMA-3-8b-odpo-v1.0 (ours)	SFT+DPO+ODPO	8B	22.0	26.8	8.19	71.41

Intended uses

To ensure reproducibility, please use the following chat templates:

python

import torch
from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model="Columbia-NLP/LION-LLaMA-3-8b-dpo-v1.0",
    device_map="auto",
    torch_dtype=torch.bfloat16,
)
messages = [
    # no system message for LLaMa
    {
        "role": "user", 
        "content": "Write a short paragraph where every sentence starts with the letter A."
    },
]
outputs = pipe(
    messages,
    max_new_tokens=128,
    do_sample=False,
    stop_sequence="<|im_end|>",
)
print(outputs[0]["generated_text"][-1]["content"])
# Astonishingly, all animals adore appealing, aromatic apples.
# An array of apples always attracts adventurous ants.
# After analyzing, ants ascertain the apple's accessibility.
# Anticipation amplifies as they approach, anticipating an appetizing treat.

to inspect the chat template/manually do generation:

python

tokenizer = AutoTokenizer.from_pretrained("Columbia-NLP/LION-LLaMA-3-8b-dpo-v1.0")
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
print(prompt)
# tokenize prompt and use model.generate

Training details

Please refer to our paper and codebase.

Citation Information

If you find this model useful in your work, please consider citing our paper:

bash

@misc{yu2024lionsempiricallyoptimizedapproach,
      title={LIONs: An Empirically Optimized Approach to Align Language Models}, 
      author={Xiao Yu and Qingyang Wu and Yu Li and Zhou Yu},
      year={2024},
      eprint={2407.06542},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2407.06542}, 
}

Acknowledgements

We thank the Columbia-NLP group and articulate.ai for providing OpenAI API credits and computational resources to conduct our experiments.

RichardErkhov/Columbia-NLP_-_LION-LLaMA-3-8b-dpo-v1.0-gguf