Jongbin-kr/llama-3.1-8b-instruct-4x1-moe-lbox-switch-top1-router-ffn-lora-aux-anneal-5ep
0213
<!-- 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. -->
llama-3.1-8b-instruct-4x1-moe-lbox-switch-top1-router-ffn-lora-aux-anneal-5ep
This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0476
- Router Entropy: 1.2596
- Router Max Load: 0.2548
- Expert 0 Utilization: 0.2527
- Expert 1 Utilization: 0.2548
- Expert 2 Utilization: 0.2522
- Expert 3 Utilization: 0.2403
- Router Aux Loss: 1.0014
- Router Aux Loss Coef: 0.0010
- Router Aux Loss Weighted: 0.0010
- Router Z Loss: 0.2350
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: 2e-05
- trainbatchsize: 1
- evalbatchsize: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradientaccumulationsteps: 16
- totaltrainbatch_size: 32
- totalevalbatch_size: 2
- optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lrschedulertype: linear
- lrschedulerwarmup_steps: 0.03
- num_epochs: 5.0
Training results
Framework versions
- PEFT 0.19.1
- Transformers 5.9.0
- Pytorch 2.11.0+cu128
- Datasets 4.8.5
- Tokenizers 0.22.2
