Hugging Face's logo

MAdAiLab
/
SLM_vs_LLM_experiments

Safetensors
Model card Files Community
SLM_vs_LLM_experiments / LoRA /Qwen /Qwen1.5_7B_LoRA_coastalcph /lex_glue_ledgar /README.md
akkky02's picture
akkky02
Upload folder using huggingface_hub
4a3dcc6 verified
preview code
|
raw
history blame
No virus
6.21 kB
metadata
license: other
library_name: peft
tags:
  - generated_from_trainer
base_model: Qwen/Qwen1.5-7B
metrics:
  - accuracy
model-index:
  - name: lex_glue_ledgar
    results: []

lex_glue_ledgar

This model is a fine-tuned version of Qwen/Qwen1.5-7B on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5025
  • Accuracy: 0.867
  • F1 Macro: 0.7910
  • F1 Micro: 0.867

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: 5e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 2
  • total_train_batch_size: 32
  • total_eval_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3.0

Training results

Training Loss Epoch Step Validation Loss Accuracy F1 Macro F1 Micro
1.7995 0.05 100 1.6894 0.6512 0.4676 0.6512
1.3922 0.11 200 1.2208 0.7076 0.5868 0.7076
1.0552 0.16 300 0.9665 0.7634 0.6329 0.7634
0.8416 0.21 400 0.9615 0.767 0.6280 0.767
0.8204 0.27 500 0.8469 0.7892 0.6680 0.7892
0.7359 0.32 600 0.7820 0.8025 0.6859 0.8025
0.7088 0.37 700 0.7905 0.7975 0.6808 0.7975
0.6096 0.43 800 0.7862 0.8009 0.6823 0.8009
0.8682 0.48 900 0.7768 0.7987 0.6967 0.7987
0.6772 0.53 1000 0.7300 0.8094 0.6934 0.8094
0.6224 0.59 1100 0.6760 0.8146 0.7190 0.8146
0.5875 0.64 1200 0.6449 0.8253 0.7442 0.8253
0.6147 0.69 1300 0.6603 0.8305 0.7208 0.8305
0.6355 0.75 1400 0.6256 0.8285 0.7294 0.8285
0.7076 0.8 1500 0.6340 0.8288 0.7290 0.8288
0.4995 0.85 1600 0.6186 0.8315 0.7422 0.8315
0.5754 0.91 1700 0.6105 0.8402 0.7482 0.8402
0.6775 0.96 1800 0.5947 0.8369 0.7531 0.8369
0.3267 1.01 1900 0.5678 0.8528 0.7704 0.8528
0.2022 1.07 2000 0.6361 0.844 0.7639 0.844
0.3831 1.12 2100 0.5957 0.8503 0.7672 0.8503
0.3235 1.17 2200 0.6062 0.8476 0.7685 0.8476
0.2279 1.23 2300 0.6255 0.847 0.7658 0.847
0.3224 1.28 2400 0.5754 0.8537 0.7772 0.8537
0.3281 1.33 2500 0.5763 0.8598 0.7769 0.8598
0.3909 1.39 2600 0.5519 0.8545 0.7778 0.8545
0.3064 1.44 2700 0.5842 0.8536 0.7790 0.8536
0.2333 1.49 2800 0.6084 0.8447 0.7674 0.8447
0.2361 1.55 2900 0.5975 0.8588 0.7853 0.8588
0.3415 1.6 3000 0.5701 0.8572 0.7844 0.8572
0.2535 1.65 3100 0.5557 0.8618 0.7828 0.8618
0.2356 1.71 3200 0.5242 0.8612 0.7822 0.8612
0.3383 1.76 3300 0.5250 0.8553 0.7873 0.8553
0.1886 1.81 3400 0.5301 0.8658 0.7924 0.8658
0.2468 1.87 3500 0.5459 0.8595 0.7813 0.8595
0.2947 1.92 3600 0.5141 0.8688 0.7910 0.8688
0.2625 1.97 3700 0.5025 0.867 0.7910 0.867
0.0829 2.03 3800 0.5625 0.8697 0.8004 0.8697
0.0297 2.08 3900 0.6303 0.8698 0.8018 0.8698
0.0474 2.13 4000 0.6244 0.8713 0.8046 0.8713
0.0267 2.19 4100 0.5801 0.8737 0.8061 0.8737
0.0487 2.24 4200 0.5915 0.8745 0.8018 0.8745
0.0272 2.29 4300 0.6174 0.8764 0.8043 0.8764
0.02 2.35 4400 0.6261 0.87 0.7986 0.87
0.0414 2.4 4500 0.6157 0.8748 0.8036 0.8748
0.0394 2.45 4600 0.6051 0.8755 0.8076 0.8755
0.0513 2.51 4700 0.6078 0.874 0.8072 0.874
0.0553 2.56 4800 0.6021 0.8734 0.8023 0.8734
0.0843 2.61 4900 0.6084 0.8766 0.8096 0.8766
0.0361 2.67 5000 0.6129 0.8764 0.8091 0.8764
0.0485 2.72 5100 0.6214 0.8789 0.8096 0.8789
0.0209 2.77 5200 0.5887 0.8795 0.8102 0.8795
0.028 2.83 5300 0.5953 0.8798 0.8132 0.8798
0.0513 2.88 5400 0.5944 0.8818 0.8154 0.8818
0.0073 2.93 5500 0.6021 0.8794 0.8136 0.8794
0.0398 2.99 5600 0.6064 0.88 0.8124 0.88

Framework versions

  • PEFT 0.9.0
  • Transformers 4.39.0.dev0
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2