CoolFace
Modelpublic

tsavage68/Na_L3_1000steps_1e6rate_03beta_cSFTDPO

sourceHugging Facellama3updated 2y agoView on Hugging Face
0likes14downloads
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. -->

NaL31000steps1e7rateSFT

This model is a fine-tuned version of tsavage68/Na_L3_100steps_1e6rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.0000
  • —Rewards/chosen: 1.8472
  • —Rewards/rejected: -12.7635
  • —Rewards/accuracies: 1.0
  • —Rewards/margins: 14.6107
  • —Logps/rejected: -84.0468
  • —Logps/chosen: -18.7329
  • —Logits/rejected: -0.9590
  • —Logits/chosen: -0.8901

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: 1000

Training results

Training LossEpochStepValidation LossRewards/chosenRewards/rejectedRewards/accuraciesRewards/marginsLogps/rejectedLogps/chosenLogits/rejectedLogits/chosen
0.00010.2667500.00011.2406-8.72741.09.9680-70.5931-20.7550-0.9609-0.8923
0.00.53331000.00001.5010-10.37161.011.8726-76.0737-19.8868-0.9605-0.8913
0.00.81500.00001.6394-11.08611.012.7255-78.4552-19.4255-0.9602-0.8910
0.01.06672000.00001.6945-11.50121.013.1957-79.8391-19.2417-0.9599-0.8908
0.01.33332500.00001.7295-11.79931.013.5288-80.8326-19.1251-0.9588-0.8896
0.01.63000.00001.7510-12.00711.013.7581-81.5254-19.0534-0.9591-0.8901
0.01.86673500.00001.7781-12.19051.013.9686-82.1367-18.9631-0.9593-0.8903
0.02.13334000.00001.7798-12.34011.014.1198-82.6353-18.9575-0.9582-0.8894
0.02.44500.00001.8074-12.46881.014.2762-83.0643-18.8654-0.9585-0.8895
0.02.66675000.00001.8066-12.56701.014.3737-83.3918-18.8680-0.9586-0.8895
0.02.93335500.00001.8171-12.63461.014.4517-83.6169-18.8330-0.9579-0.8889
0.03.26000.00001.8284-12.68061.014.5090-83.7705-18.7955-0.9589-0.8901
0.03.46676500.00001.8462-12.72641.014.5726-83.9231-18.7362-0.9585-0.8897
0.03.73337000.00001.8460-12.75601.014.6020-84.0217-18.7367-0.9589-0.8899
0.04.07500.00001.8460-12.75581.014.6019-84.0211-18.7366-0.9588-0.8899
0.04.26678000.00001.8430-12.76831.014.6113-84.0628-18.7468-0.9578-0.8889
0.04.53338500.00001.8450-12.77751.014.6226-84.0936-18.7401-0.9586-0.8898
0.04.89000.00001.8391-12.76341.014.6025-84.0465-18.7599-0.9587-0.8898
0.05.06679500.00001.8471-12.76491.014.6120-84.0513-18.7330-0.9590-0.8901
0.05.333310000.00001.8472-12.76351.014.6107-84.0468-18.7329-0.9590-0.8901

Framework versions

  • —Transformers 4.44.2
  • —Pytorch 2.4.0+cu121
  • —Datasets 2.21.0
  • —Tokenizers 0.19.1