CoolFace
Modelpublic

RichardErkhov/tsavage68_-_Na_M2_1000steps_1e8rate_03beta_cSFTDPO-gguf

sourceHugging Faceupdated 1y agoView on Hugging Face
0likes225downloads
Model Card

Quantization made by Richard Erkhov.

Github

Discord

Request more models

NaM21000steps1e8rate03beta_cSFTDPO - GGUF

  • —Model creator: https://huggingface.co/tsavage68/
  • —Original model: https://huggingface.co/tsavage68/NaM21000steps1e8rate03beta_cSFTDPO/
NameQuant methodSize
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q2_K.ggufQ2_K2.53GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.IQ3_XS.ggufIQ3_XS2.81GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.IQ3_S.ggufIQ3_S2.96GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q3_K_S.ggufQ3KS2.95GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.IQ3_M.ggufIQ3_M3.06GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q3_K.ggufQ3_K3.28GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q3_K_M.ggufQ3KM3.28GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q3_K_L.ggufQ3KL3.56GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.IQ4_XS.ggufIQ4_XS3.67GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q4_0.ggufQ4_03.83GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.IQ4_NL.ggufIQ4_NL3.87GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q4_K_S.ggufQ4KS3.86GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q4_K.ggufQ4_K4.07GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q4_K_M.ggufQ4KM4.07GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q4_1.ggufQ4_14.24GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q5_0.ggufQ5_04.65GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q5_K_S.ggufQ5KS4.65GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q5_K.ggufQ5_K4.78GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q5_K_M.ggufQ5KM4.78GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q5_1.ggufQ5_15.07GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q6_K.ggufQ6_K5.53GB
Na_M2_1000steps_1e8rate_03beta_cSFTDPO.Q8_0.ggufQ8_07.17GB

Original model description: --- libraryname: transformers license: apache-2.0 basemodel: tsavage68/NaM21000steps1e7SFT tags:

  • —trl
  • —dpo
  • —generatedfromtrainer model-index:
  • —name: NaM21000steps1e8rate03beta_cSFTDPO results: [] ---

<!-- 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. -->

NaM21000steps1e8rate03beta_cSFTDPO

This model is a fine-tuned version of tsavage68/Na_M2_1000steps_1e7_SFT on an unknown dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.4450
  • —Rewards/chosen: 0.1680
  • —Rewards/rejected: -0.4255
  • —Rewards/accuracies: 1.0
  • —Rewards/margins: 0.5934
  • —Logps/rejected: -81.3416
  • —Logps/chosen: -47.5724
  • —Logits/rejected: -2.5355
  • —Logits/chosen: -2.5481

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-08
  • —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
0.69550.2667500.68820.0099-0.00310.56000.0130-79.9338-48.0995-2.5354-2.5481
0.67610.53331000.67300.0130-0.03150.66000.0445-80.0283-48.0889-2.5363-2.5489
0.61540.81500.59710.0672-0.13930.98000.2065-80.3878-47.9083-2.5367-2.5493
0.57351.06672000.54300.1029-0.23021.00.3331-80.6906-47.7893-2.5352-2.5478
0.50471.33332500.50200.1363-0.30301.00.4393-80.9334-47.6779-2.5353-2.5478
0.45251.63000.47510.1411-0.36851.00.5096-81.1517-47.6622-2.5350-2.5476
0.4511.86673500.45720.1576-0.39881.00.5564-81.2528-47.6072-2.5350-2.5475
0.44342.13334000.45010.1391-0.43871.00.5778-81.3857-47.6686-2.5351-2.5477
0.43132.44500.44540.1528-0.43701.00.5899-81.3802-47.6230-2.5343-2.5469
0.45462.66675000.45130.1462-0.42931.00.5755-81.3544-47.6450-2.5345-2.5471
0.45262.93335500.44240.1917-0.41101.00.6027-81.2934-47.4934-2.5352-2.5476
0.44263.26000.44370.1805-0.41751.00.5980-81.3150-47.5307-2.5361-2.5486
0.44523.46676500.44030.1651-0.43921.00.6043-81.3875-47.5821-2.5347-2.5473
0.4183.73337000.44500.1668-0.42371.00.5905-81.3358-47.5764-2.5348-2.5474
0.42814.07500.44500.1680-0.42551.00.5934-81.3416-47.5724-2.5355-2.5481
0.45034.26678000.44500.1680-0.42551.00.5934-81.3416-47.5724-2.5355-2.5481
0.43724.53338500.44500.1680-0.42551.00.5934-81.3416-47.5724-2.5355-2.5481
0.41354.89000.44500.1680-0.42551.00.5934-81.3416-47.5724-2.5355-2.5481
0.43165.06679500.44500.1680-0.42551.00.5934-81.3416-47.5724-2.5355-2.5481
0.44385.333310000.44500.1680-0.42551.00.5934-81.3416-47.5724-2.5355-2.5481

Framework versions

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