metadata
library_name: peft
license: other
base_model: >-
/home/aiops/wanghy/models/hub/models--GraySwanAI--Llama-3-8B-Instruct-RR/snapshots/d92f951d380d3489fb56b08c296376ea61cebef
tags:
- llama-factory
- lora
- generated_from_trainer
model-index:
- name: train_RR_i2_sametime_jailbreak_math10k_simple
results: []
train_RR_i2_sametime_jailbreak_math10k_simple
This model is a fine-tuned version of /home/aiops/wanghy/models/hub/models--GraySwanAI--Llama-3-8B-Instruct-RR/snapshots/d92f951d380d3489fb56b08c296376ea61cebef on the train_metamath10k, the train_jailbreakchat, the train_RR_i1, the train_RR_i2 and the train_simple_adaptive_attack datasets. It achieves the following results on the evaluation set:
- Loss: 0.3195
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: 0.0001
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 64
- total_eval_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.5133 | 0.6327 | 500 | 0.2772 |
| 0.4751 | 1.2654 | 1000 | 0.2744 |
| 0.4715 | 1.8981 | 1500 | 0.2363 |
| 0.4275 | 2.5308 | 2000 | 0.2949 |
Framework versions
- PEFT 0.12.0
- Transformers 4.45.0
- Pytorch 2.4.0+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3