--- license: mit library_name: peft tags: - generated_from_trainer base_model: microsoft/phi-2 model-index: - name: phi-2-hummanize1 results: [] --- # phi-2-hummanize1 This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co./microsoft/phi-2) on the None dataset. It achieves the following results on the evaluation set: - Loss: 6.7074 ## 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: 0.0025 - train_batch_size: 20 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 20 - total_train_batch_size: 400 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 2 - num_epochs: 2 ### Training results | Training Loss | Epoch | Step | Validation Loss | |:-------------:|:-----:|:----:|:---------------:| | No log | 0.02 | 1 | 2.0864 | | No log | 0.05 | 2 | 4.0671 | | No log | 0.07 | 3 | 3.6332 | | No log | 0.09 | 4 | 2.5537 | | 3.0197 | 0.11 | 5 | 2.3394 | | 3.0197 | 0.14 | 6 | 2.8862 | | 3.0197 | 0.16 | 7 | 2.5140 | | 3.0197 | 0.18 | 8 | 2.4603 | | 3.0197 | 0.21 | 9 | 2.2094 | | 2.5958 | 0.23 | 10 | 2.1767 | | 2.5958 | 0.25 | 11 | 2.3343 | | 2.5958 | 0.28 | 12 | 2.2511 | | 2.5958 | 0.3 | 13 | 2.1854 | | 2.5958 | 0.32 | 14 | 2.1385 | | 2.2944 | 0.34 | 15 | 2.3556 | | 2.2944 | 0.37 | 16 | 2.2056 | | 2.2944 | 0.39 | 17 | 2.2127 | | 2.2944 | 0.41 | 18 | 2.1507 | | 2.2944 | 0.44 | 19 | 2.1388 | | 2.2841 | 0.46 | 20 | 2.6540 | | 2.2841 | 0.48 | 21 | 2.8934 | | 2.2841 | 0.51 | 22 | 3.0981 | | 2.2841 | 0.53 | 23 | 2.4155 | | 2.2841 | 0.55 | 24 | 2.1754 | | 2.7585 | 0.57 | 25 | 2.0927 | | 2.7585 | 0.6 | 26 | 2.0865 | | 2.7585 | 0.62 | 27 | 2.2345 | | 2.7585 | 0.64 | 28 | 2.4123 | | 2.7585 | 0.67 | 29 | 2.7718 | | 2.3906 | 0.69 | 30 | 4.2964 | | 2.3906 | 0.71 | 31 | 6.5295 | | 2.3906 | 0.73 | 32 | 5.8489 | | 2.3906 | 0.76 | 33 | 7.2467 | | 2.3906 | 0.78 | 34 | 7.6353 | | 6.5839 | 0.8 | 35 | 7.7842 | | 6.5839 | 0.83 | 36 | 8.8627 | | 6.5839 | 0.85 | 37 | 7.9511 | | 6.5839 | 0.87 | 38 | 9.7736 | | 6.5839 | 0.9 | 39 | 8.3666 | | 8.6795 | 0.92 | 40 | 8.9768 | | 8.6795 | 0.94 | 41 | 9.0808 | | 8.6795 | 0.96 | 42 | 8.5933 | | 8.6795 | 0.99 | 43 | 8.9317 | | 8.6795 | 1.01 | 44 | 8.5291 | | 8.8177 | 1.03 | 45 | 8.5935 | | 8.8177 | 1.06 | 46 | 8.6773 | | 8.8177 | 1.08 | 47 | 8.5914 | | 8.8177 | 1.1 | 48 | 8.5006 | | 8.8177 | 1.13 | 49 | 8.3959 | | 8.6883 | 1.15 | 50 | 8.2375 | | 8.6883 | 1.17 | 51 | 8.2022 | | 8.6883 | 1.19 | 52 | 8.2063 | | 8.6883 | 1.22 | 53 | 8.2254 | | 8.6883 | 1.24 | 54 | 8.3408 | | 8.4216 | 1.26 | 55 | 8.0367 | | 8.4216 | 1.29 | 56 | 7.8776 | | 8.4216 | 1.31 | 57 | 7.6720 | | 8.4216 | 1.33 | 58 | 7.5050 | | 8.4216 | 1.35 | 59 | 7.3863 | | 7.8151 | 1.38 | 60 | 7.3775 | | 7.8151 | 1.4 | 61 | 7.3820 | | 7.8151 | 1.42 | 62 | 7.2597 | | 7.8151 | 1.45 | 63 | 7.1959 | | 7.8151 | 1.47 | 64 | 7.1233 | | 7.3639 | 1.49 | 65 | 7.0625 | | 7.3639 | 1.52 | 66 | 7.0302 | | 7.3639 | 1.54 | 67 | 6.9862 | | 7.3639 | 1.56 | 68 | 6.9601 | | 7.3639 | 1.58 | 69 | 6.9606 | | 7.1152 | 1.61 | 70 | 6.8977 | | 7.1152 | 1.63 | 71 | 6.8981 | | 7.1152 | 1.65 | 72 | 6.8453 | | 7.1152 | 1.68 | 73 | 6.8523 | | 7.1152 | 1.7 | 74 | 6.8641 | | 6.9712 | 1.72 | 75 | 6.8261 | | 6.9712 | 1.75 | 76 | 6.8273 | | 6.9712 | 1.77 | 77 | 6.8053 | | 6.9712 | 1.79 | 78 | 6.7712 | | 6.9712 | 1.81 | 79 | 6.7542 | | 6.8925 | 1.84 | 80 | 6.7466 | | 6.8925 | 1.86 | 81 | 6.7341 | | 6.8925 | 1.88 | 82 | 6.7255 | | 6.8925 | 1.91 | 83 | 6.7211 | | 6.8925 | 1.93 | 84 | 6.7154 | | 6.8192 | 1.95 | 85 | 6.7103 | | 6.8192 | 1.97 | 86 | 6.7074 | ### Framework versions - PEFT 0.8.2 - Transformers 4.38.0.dev0 - Pytorch 2.1.0+cu118 - Datasets 2.17.0 - Tokenizers 0.15.2