CoolFace
Modelpublic

tsavage68/Transaminitis_L3_100steps_1e6rate_01beta_CSFTDPO

sourceHugging Facellama3updated 2y agoView on Hugging Face
0likes12downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

TransaminitisL3100steps1e6rate01beta_CSFTDPO

This model is a fine-tuned version of tsavage68/Transaminitis_L3_1000rate_1e7_SFT on an unknown dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.0003
  • —Rewards/chosen: 1.7922
  • —Rewards/rejected: -6.4799
  • —Rewards/accuracies: 1.0
  • —Rewards/margins: 8.2721
  • —Logps/rejected: -83.3533
  • —Logps/chosen: -0.6118
  • —Logits/rejected: -0.9682
  • —Logits/chosen: -0.9313

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • —learning_rate: 1e-06
  • —trainbatchsize: 2
  • —evalbatchsize: 1
  • —seed: 42
  • —gradientaccumulationsteps: 2
  • —totaltrainbatch_size: 4
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: cosine
  • —lrschedulerwarmup_steps: 100
  • —training_steps: 100

Training results

Training LossEpochStepValidation LossRewards/chosenRewards/rejectedRewards/accuraciesRewards/marginsLogps/rejectedLogps/chosenLogits/rejectedLogits/chosen
0.69250.2250.6848-0.0689-0.08600.72000.0171-19.4148-19.2230-1.0686-1.0674
0.64110.4500.45660.6281-0.01040.87000.6385-18.6588-12.2529-1.0509-1.0497
0.19780.6750.28041.4640-5.94580.95007.4098-78.0130-3.8943-0.9527-0.9153
0.4320.81000.00031.7922-6.47991.08.2721-83.3533-0.6118-0.9682-0.9313

Framework versions

  • —Transformers 4.40.2
  • —Pytorch 2.0.0+cu117
  • —Datasets 2.19.1
  • —Tokenizers 0.19.1