tsavage68/Na_L3_1000steps_1e6rate_03beta_cSFTDPO
014
<!-- 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
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu121
- Datasets 2.21.0
- Tokenizers 0.19.1
