RichardErkhov/qgallouedec_-_online-dpo-qwen2-4-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

online-dpo-qwen2-4 - GGUF

• Model creator: https://huggingface.co/qgallouedec/
• Original model: https://huggingface.co/qgallouedec/online-dpo-qwen2-4/

Name	Quant method	Size
online-dpo-qwen2-4.Q2_K.gguf	Q2_K	0.32GB
online-dpo-qwen2-4.IQ3_XS.gguf	IQ3_XS	0.32GB
online-dpo-qwen2-4.IQ3_S.gguf	IQ3_S	0.32GB
online-dpo-qwen2-4.Q3_K_S.gguf	Q3_K_S	0.32GB
online-dpo-qwen2-4.IQ3_M.gguf	IQ3_M	0.32GB
online-dpo-qwen2-4.Q3_K.gguf	Q3_K	0.33GB
online-dpo-qwen2-4.Q3_K_M.gguf	Q3_K_M	0.33GB
online-dpo-qwen2-4.Q3_K_L.gguf	Q3_K_L	0.34GB
online-dpo-qwen2-4.IQ4_XS.gguf	IQ4_XS	0.33GB
online-dpo-qwen2-4.Q4_0.gguf	Q4_0	0.33GB
online-dpo-qwen2-4.IQ4_NL.gguf	IQ4_NL	0.33GB
online-dpo-qwen2-4.Q4_K_S.gguf	Q4_K_S	0.36GB
online-dpo-qwen2-4.Q4_K.gguf	Q4_K	0.37GB
online-dpo-qwen2-4.Q4_K_M.gguf	Q4_K_M	0.37GB
online-dpo-qwen2-4.Q4_1.gguf	Q4_1	0.35GB
online-dpo-qwen2-4.Q5_0.gguf	Q5_0	0.37GB
online-dpo-qwen2-4.Q5_K_S.gguf	Q5_K_S	0.38GB
online-dpo-qwen2-4.Q5_K.gguf	Q5_K	0.39GB
online-dpo-qwen2-4.Q5_K_M.gguf	Q5_K_M	0.39GB
online-dpo-qwen2-4.Q5_1.gguf	Q5_1	0.39GB
online-dpo-qwen2-4.Q6_K.gguf	Q6_K	0.47GB
online-dpo-qwen2-4.Q8_0.gguf	Q8_0	0.49GB

Original model description:

base_model: Qwen/Qwen2-0.5B-Instruct datasets: trl-lib/ultrafeedback-prompt library_name: transformers model_name: online-dpo-qwen2-4 tags: - trl - generated_from_trainer - online-dpo licence: license

Model Card for online-dpo-qwen2-4

This model is a fine-tuned version of Qwen/Qwen2-0.5B-Instruct on the trl-lib/ultrafeedback-prompt dataset. It has been trained using TRL.

Quick start

from transformers import pipeline

question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="qgallouedec/online-dpo-qwen2-4", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=500)[0]
print(output["generated_text"][1]["content"])

Training procedure

This model was trained with Online DPO, a method introduced in Direct Language Model Alignment from Online AI Feedback.

Framework versions

• TRL: 0.12.0.dev0
• Transformers: 4.45.0.dev0
• Pytorch: 2.4.1
• Datasets: 3.0.0
• Tokenizers: 0.19.1