You can deploy this model and start earning money today!

RichardErkhov/
OpenRLHF_-_Llama-3-8b-iter-dpo-179k-gguf

Name: OpenRLHF_-_Llama-3-8b-iter-dpo-179k-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Llama-3-8b-iter-dpo-179k - GGUF

Model creator: https://huggingface.co/OpenRLHF/
Original model: https://huggingface.co/OpenRLHF/Llama-3-8b-iter-dpo-179k/

Name	Quant method	Size
Llama-3-8b-iter-dpo-179k.Q2_K.gguf	Q2_K	2.96GB
Llama-3-8b-iter-dpo-179k.IQ3_XS.gguf	IQ3_XS	3.28GB
Llama-3-8b-iter-dpo-179k.IQ3_S.gguf	IQ3_S	3.43GB
Llama-3-8b-iter-dpo-179k.Q3_K_S.gguf	Q3_K_S	3.41GB
Llama-3-8b-iter-dpo-179k.IQ3_M.gguf	IQ3_M	3.52GB
Llama-3-8b-iter-dpo-179k.Q3_K.gguf	Q3_K	3.74GB
Llama-3-8b-iter-dpo-179k.Q3_K_M.gguf	Q3_K_M	3.74GB
Llama-3-8b-iter-dpo-179k.Q3_K_L.gguf	Q3_K_L	4.03GB
Llama-3-8b-iter-dpo-179k.IQ4_XS.gguf	IQ4_XS	4.18GB
Llama-3-8b-iter-dpo-179k.Q4_0.gguf	Q4_0	4.34GB
Llama-3-8b-iter-dpo-179k.IQ4_NL.gguf	IQ4_NL	4.38GB
Llama-3-8b-iter-dpo-179k.Q4_K_S.gguf	Q4_K_S	4.37GB
Llama-3-8b-iter-dpo-179k.Q4_K.gguf	Q4_K	4.58GB
Llama-3-8b-iter-dpo-179k.Q4_K_M.gguf	Q4_K_M	4.58GB
Llama-3-8b-iter-dpo-179k.Q4_1.gguf	Q4_1	4.78GB
Llama-3-8b-iter-dpo-179k.Q5_0.gguf	Q5_0	5.21GB
Llama-3-8b-iter-dpo-179k.Q5_K_S.gguf	Q5_K_S	5.21GB
Llama-3-8b-iter-dpo-179k.Q5_K.gguf	Q5_K	5.34GB
Llama-3-8b-iter-dpo-179k.Q5_K_M.gguf	Q5_K_M	5.34GB
Llama-3-8b-iter-dpo-179k.Q5_1.gguf	Q5_1	5.65GB
Llama-3-8b-iter-dpo-179k.Q6_K.gguf	Q6_K	6.14GB
Llama-3-8b-iter-dpo-179k.Q8_0.gguf	Q8_0	7.95GB

Original model description: This model is trained with Iterative DPO in OpenRLHF

Datasets and Hyperparameters

Reward Model:https://huggingface.co/OpenLLMAI/Llama-3-8b-rm-700k
SFT Model: https://huggingface.co/OpenLLMAI/Llama-3-8b-sft-mixture
Prompt Dataset: https://huggingface.co/datasets/OpenLLMAI/prompt-collection-v0.1

Max Prompt Length: 2048
Max Response Length: 2048
best_of_n: 2 (2 samples for each prompt)
Learning Rate: 5e-7
Beta: 0.1
Scheduler: Cosine with Warmup (0.03) and MinLR (0.1 * init_lr)
Rollout Batch Size: 20000
Training Batch Size: 256
Number of Iterations: 9

Evaluation

########## First turn ##########
                      score
model           turn
Llama3-iter-dpo 1      8.55
########## Second turn ##########
                        score
model           turn
Llama3-iter-dpo 2     7.95625
########## Average ##########
                    score
model
Llama3-iter-dpo  8.253125
Llama3-sft-baseline 7.69

RichardErkhov/OpenRLHF_-_Llama-3-8b-iter-dpo-179k-gguf

RichardErkhov/
OpenRLHF_-_Llama-3-8b-iter-dpo-179k-gguf