| [hops] 2024-09-23 22:03:10.261 | INFO | Initializing a parser from /workspace/configs/exp_camembertv2/camembertv2_base_p2_17k_last_layer.yaml | |
| [hops] 2024-09-23 22:03:10.554 | INFO | Generating a FastText model from the treebank | |
| [hops] 2024-09-23 22:03:10.645 | INFO | Training fasttext model | |
| [hops] 2024-09-23 22:03:12.421 | WARNING | Some weights of RobertaModel were not initialized from the model checkpoint at /scratch/camembertv2/runs/models/camembertv2-base-bf16/post/ckpt-p2-17000/pt/ and are newly initialized: ['roberta.pooler.dense.bias', 'roberta.pooler.dense.weight'] | |
| You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. | |
| [hops] 2024-09-23 22:03:24.938 | INFO | Start training on cuda:0 | |
| [hops] 2024-09-23 22:03:24.944 | WARNING | You're using a RobertaTokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. | |
| [hops] 2024-09-23 22:04:54.185 | INFO | Epoch 0: train loss 1.1740 dev loss 0.3235 dev tag acc 93.16% dev head acc 88.90% dev deprel acc 91.95% | |
| [hops] 2024-09-23 22:04:54.186 | INFO | New best model: head accuracy 88.90% > 0.00% | |
| [hops] 2024-09-23 22:06:23.342 | INFO | Epoch 1: train loss 0.2838 dev loss 0.1626 dev tag acc 97.85% dev head acc 94.35% dev deprel acc 95.43% | |
| [hops] 2024-09-23 22:06:23.343 | INFO | New best model: head accuracy 94.35% > 88.90% | |
| [hops] 2024-09-23 22:07:57.103 | INFO | Epoch 2: train loss 0.1673 dev loss 0.1337 dev tag acc 98.24% dev head acc 95.40% dev deprel acc 96.71% | |
| [hops] 2024-09-23 22:07:57.104 | INFO | New best model: head accuracy 95.40% > 94.35% | |
| [hops] 2024-09-23 22:09:24.832 | INFO | Epoch 3: train loss 0.1227 dev loss 0.1438 dev tag acc 98.43% dev head acc 95.90% dev deprel acc 96.81% | |
| [hops] 2024-09-23 22:09:24.833 | INFO | New best model: head accuracy 95.90% > 95.40% | |
| [hops] 2024-09-23 22:10:55.835 | INFO | Epoch 4: train loss 0.0968 dev loss 0.1418 dev tag acc 98.43% dev head acc 96.14% dev deprel acc 97.20% | |
| [hops] 2024-09-23 22:10:55.836 | INFO | New best model: head accuracy 96.14% > 95.90% | |
| [hops] 2024-09-23 22:12:27.472 | INFO | Epoch 5: train loss 0.0793 dev loss 0.1568 dev tag acc 98.52% dev head acc 96.19% dev deprel acc 97.34% | |
| [hops] 2024-09-23 22:12:27.473 | INFO | New best model: head accuracy 96.19% > 96.14% | |
| [hops] 2024-09-23 22:13:56.231 | INFO | Epoch 6: train loss 0.0667 dev loss 0.1516 dev tag acc 98.57% dev head acc 96.40% dev deprel acc 97.36% | |
| [hops] 2024-09-23 22:13:56.232 | INFO | New best model: head accuracy 96.40% > 96.19% | |
| [hops] 2024-09-23 22:15:26.147 | INFO | Epoch 7: train loss 0.0566 dev loss 0.1687 dev tag acc 98.58% dev head acc 96.52% dev deprel acc 97.52% | |
| [hops] 2024-09-23 22:15:26.148 | INFO | New best model: head accuracy 96.52% > 96.40% | |
| [hops] 2024-09-23 22:16:55.419 | INFO | Epoch 8: train loss 0.0500 dev loss 0.1826 dev tag acc 98.64% dev head acc 96.53% dev deprel acc 97.42% | |
| [hops] 2024-09-23 22:16:55.420 | INFO | New best model: head accuracy 96.53% > 96.52% | |
| [hops] 2024-09-23 22:18:28.485 | INFO | Epoch 9: train loss 0.0425 dev loss 0.1906 dev tag acc 98.59% dev head acc 96.56% dev deprel acc 97.56% | |
| [hops] 2024-09-23 22:18:28.486 | INFO | New best model: head accuracy 96.56% > 96.53% | |
| [hops] 2024-09-23 22:20:02.390 | INFO | Epoch 10: train loss 0.0377 dev loss 0.2151 dev tag acc 98.55% dev head acc 96.60% dev deprel acc 97.51% | |
| [hops] 2024-09-23 22:20:02.390 | INFO | New best model: head accuracy 96.60% > 96.56% | |
| [hops] 2024-09-23 22:21:31.071 | INFO | Epoch 11: train loss 0.0332 dev loss 0.2276 dev tag acc 98.59% dev head acc 96.62% dev deprel acc 97.53% | |
| [hops] 2024-09-23 22:21:31.072 | INFO | New best model: head accuracy 96.62% > 96.60% | |
| [hops] 2024-09-23 22:22:58.744 | INFO | Epoch 12: train loss 0.0299 dev loss 0.2397 dev tag acc 98.59% dev head acc 96.62% dev deprel acc 97.49% | |
| [hops] 2024-09-23 22:22:58.745 | INFO | New best model: head accuracy 96.62% > 96.62% | |
| [hops] 2024-09-23 22:24:30.195 | INFO | Epoch 13: train loss 0.0270 dev loss 0.2548 dev tag acc 98.64% dev head acc 96.45% dev deprel acc 97.61% | |
| [hops] 2024-09-23 22:25:58.937 | INFO | Epoch 14: train loss 0.0247 dev loss 0.2351 dev tag acc 98.69% dev head acc 96.51% dev deprel acc 97.60% | |
| [hops] 2024-09-23 22:27:26.485 | INFO | Epoch 15: train loss 0.0219 dev loss 0.2812 dev tag acc 98.64% dev head acc 96.60% dev deprel acc 97.63% | |
| [hops] 2024-09-23 22:28:53.871 | INFO | Epoch 16: train loss 0.0204 dev loss 0.2771 dev tag acc 98.64% dev head acc 96.70% dev deprel acc 97.59% | |
| [hops] 2024-09-23 22:28:53.872 | INFO | New best model: head accuracy 96.70% > 96.62% | |
| [hops] 2024-09-23 22:30:22.009 | INFO | Epoch 17: train loss 0.0193 dev loss 0.2966 dev tag acc 98.57% dev head acc 96.71% dev deprel acc 97.54% | |
| [hops] 2024-09-23 22:30:22.010 | INFO | New best model: head accuracy 96.71% > 96.70% | |
| [hops] 2024-09-23 22:31:50.178 | INFO | Epoch 18: train loss 0.0172 dev loss 0.3181 dev tag acc 98.65% dev head acc 96.63% dev deprel acc 97.61% | |
| [hops] 2024-09-23 22:33:18.205 | INFO | Epoch 19: train loss 0.0163 dev loss 0.3030 dev tag acc 98.66% dev head acc 96.73% dev deprel acc 97.62% | |
| [hops] 2024-09-23 22:33:18.206 | INFO | New best model: head accuracy 96.73% > 96.71% | |
| [hops] 2024-09-23 22:34:52.436 | INFO | Epoch 20: train loss 0.0150 dev loss 0.3732 dev tag acc 98.64% dev head acc 96.74% dev deprel acc 97.44% | |
| [hops] 2024-09-23 22:34:52.437 | INFO | New best model: head accuracy 96.74% > 96.73% | |
| [hops] 2024-09-23 22:36:26.028 | INFO | Epoch 21: train loss 0.0139 dev loss 0.3404 dev tag acc 98.59% dev head acc 96.74% dev deprel acc 97.57% | |
| [hops] 2024-09-23 22:36:26.029 | INFO | New best model: head accuracy 96.74% > 96.74% | |
| [hops] 2024-09-23 22:37:56.614 | INFO | Epoch 22: train loss 0.0130 dev loss 0.3795 dev tag acc 98.66% dev head acc 96.59% dev deprel acc 97.59% | |
| [hops] 2024-09-23 22:39:24.997 | INFO | Epoch 23: train loss 0.0120 dev loss 0.3572 dev tag acc 98.70% dev head acc 96.67% dev deprel acc 97.71% | |
| [hops] 2024-09-23 22:40:54.945 | INFO | Epoch 24: train loss 0.0114 dev loss 0.3795 dev tag acc 98.65% dev head acc 96.71% dev deprel acc 97.69% | |
| [hops] 2024-09-23 22:42:25.287 | INFO | Epoch 25: train loss 0.0113 dev loss 0.3792 dev tag acc 98.57% dev head acc 96.60% dev deprel acc 97.59% | |
| [hops] 2024-09-23 22:43:52.396 | INFO | Epoch 26: train loss 0.0105 dev loss 0.3807 dev tag acc 98.69% dev head acc 96.61% dev deprel acc 97.63% | |
| [hops] 2024-09-23 22:45:20.429 | INFO | Epoch 27: train loss 0.0093 dev loss 0.4159 dev tag acc 98.66% dev head acc 96.71% dev deprel acc 97.65% | |
| [hops] 2024-09-23 22:46:51.804 | INFO | Epoch 28: train loss 0.0088 dev loss 0.4024 dev tag acc 98.56% dev head acc 96.68% dev deprel acc 97.59% | |
| [hops] 2024-09-23 22:48:21.306 | INFO | Epoch 29: train loss 0.0084 dev loss 0.4070 dev tag acc 98.58% dev head acc 96.69% dev deprel acc 97.66% | |
| [hops] 2024-09-23 22:49:52.685 | INFO | Epoch 30: train loss 0.0085 dev loss 0.4418 dev tag acc 98.58% dev head acc 96.70% dev deprel acc 97.64% | |
| [hops] 2024-09-23 22:51:21.719 | INFO | Epoch 31: train loss 0.0077 dev loss 0.4297 dev tag acc 98.62% dev head acc 96.67% dev deprel acc 97.66% | |
| [hops] 2024-09-23 22:52:56.380 | INFO | Epoch 32: train loss 0.0070 dev loss 0.4392 dev tag acc 98.63% dev head acc 96.63% dev deprel acc 97.71% | |
| [hops] 2024-09-23 22:54:24.344 | INFO | Epoch 33: train loss 0.0065 dev loss 0.5069 dev tag acc 98.69% dev head acc 96.65% dev deprel acc 97.61% | |
| [hops] 2024-09-23 22:55:56.289 | INFO | Epoch 34: train loss 0.0066 dev loss 0.4738 dev tag acc 98.64% dev head acc 96.57% dev deprel acc 97.58% | |
| [hops] 2024-09-23 22:57:26.001 | INFO | Epoch 35: train loss 0.0059 dev loss 0.4935 dev tag acc 98.60% dev head acc 96.62% dev deprel acc 97.57% | |
| [hops] 2024-09-23 22:58:52.412 | INFO | Epoch 36: train loss 0.0056 dev loss 0.5007 dev tag acc 98.65% dev head acc 96.57% dev deprel acc 97.55% | |
| [hops] 2024-09-23 23:00:21.973 | INFO | Epoch 37: train loss 0.0053 dev loss 0.5094 dev tag acc 98.60% dev head acc 96.71% dev deprel acc 97.54% | |
| [hops] 2024-09-23 23:01:50.675 | INFO | Epoch 38: train loss 0.0051 dev loss 0.4747 dev tag acc 98.61% dev head acc 96.73% dev deprel acc 97.57% | |
| [hops] 2024-09-23 23:03:21.971 | INFO | Epoch 39: train loss 0.0048 dev loss 0.5596 dev tag acc 98.65% dev head acc 96.73% dev deprel acc 97.65% | |
| [hops] 2024-09-23 23:04:50.664 | INFO | Epoch 40: train loss 0.0043 dev loss 0.4880 dev tag acc 98.67% dev head acc 96.79% dev deprel acc 97.69% | |
| [hops] 2024-09-23 23:04:50.665 | INFO | New best model: head accuracy 96.79% > 96.74% | |
| [hops] 2024-09-23 23:06:18.898 | INFO | Epoch 41: train loss 0.0041 dev loss 0.5152 dev tag acc 98.69% dev head acc 96.68% dev deprel acc 97.65% | |
| [hops] 2024-09-23 23:07:48.810 | INFO | Epoch 42: train loss 0.0042 dev loss 0.5796 dev tag acc 98.62% dev head acc 96.77% dev deprel acc 97.59% | |
| [hops] 2024-09-23 23:09:19.338 | INFO | Epoch 43: train loss 0.0039 dev loss 0.5478 dev tag acc 98.66% dev head acc 96.69% dev deprel acc 97.69% | |
| [hops] 2024-09-23 23:10:49.453 | INFO | Epoch 44: train loss 0.0034 dev loss 0.5761 dev tag acc 98.66% dev head acc 96.71% dev deprel acc 97.64% | |
| [hops] 2024-09-23 23:12:20.508 | INFO | Epoch 45: train loss 0.0035 dev loss 0.5968 dev tag acc 98.64% dev head acc 96.75% dev deprel acc 97.63% | |
| [hops] 2024-09-23 23:13:49.566 | INFO | Epoch 46: train loss 0.0032 dev loss 0.5657 dev tag acc 98.64% dev head acc 96.78% dev deprel acc 97.69% | |
| [hops] 2024-09-23 23:15:19.300 | INFO | Epoch 47: train loss 0.0029 dev loss 0.6033 dev tag acc 98.67% dev head acc 96.72% dev deprel acc 97.68% | |
| [hops] 2024-09-23 23:16:46.854 | INFO | Epoch 48: train loss 0.0029 dev loss 0.6110 dev tag acc 98.67% dev head acc 96.72% dev deprel acc 97.68% | |
| [hops] 2024-09-23 23:18:14.068 | INFO | Epoch 49: train loss 0.0026 dev loss 0.6084 dev tag acc 98.68% dev head acc 96.74% dev deprel acc 97.67% | |
| [hops] 2024-09-23 23:19:44.122 | INFO | Epoch 50: train loss 0.0025 dev loss 0.6095 dev tag acc 98.62% dev head acc 96.76% dev deprel acc 97.68% | |
| [hops] 2024-09-23 23:21:10.264 | INFO | Epoch 51: train loss 0.0025 dev loss 0.6551 dev tag acc 98.69% dev head acc 96.71% dev deprel acc 97.73% | |
| [hops] 2024-09-23 23:22:41.212 | INFO | Epoch 52: train loss 0.0022 dev loss 0.6374 dev tag acc 98.62% dev head acc 96.70% dev deprel acc 97.64% | |
| [hops] 2024-09-23 23:24:09.182 | INFO | Epoch 53: train loss 0.0021 dev loss 0.6473 dev tag acc 98.64% dev head acc 96.72% dev deprel acc 97.64% | |
| [hops] 2024-09-23 23:25:37.902 | INFO | Epoch 54: train loss 0.0019 dev loss 0.6793 dev tag acc 98.66% dev head acc 96.73% dev deprel acc 97.67% | |
| [hops] 2024-09-23 23:27:06.796 | INFO | Epoch 55: train loss 0.0019 dev loss 0.6544 dev tag acc 98.66% dev head acc 96.76% dev deprel acc 97.70% | |
| [hops] 2024-09-23 23:28:34.813 | INFO | Epoch 56: train loss 0.0016 dev loss 0.7122 dev tag acc 98.66% dev head acc 96.69% dev deprel acc 97.67% | |
| [hops] 2024-09-23 23:30:01.304 | INFO | Epoch 57: train loss 0.0015 dev loss 0.7413 dev tag acc 98.65% dev head acc 96.69% dev deprel acc 97.68% | |
| [hops] 2024-09-23 23:31:30.980 | INFO | Epoch 58: train loss 0.0015 dev loss 0.7386 dev tag acc 98.66% dev head acc 96.71% dev deprel acc 97.68% | |
| [hops] 2024-09-23 23:33:00.851 | INFO | Epoch 59: train loss 0.0014 dev loss 0.7433 dev tag acc 98.65% dev head acc 96.80% dev deprel acc 97.68% | |
| [hops] 2024-09-23 23:33:00.852 | INFO | New best model: head accuracy 96.80% > 96.79% | |
| [hops] 2024-09-23 23:34:32.878 | INFO | Epoch 60: train loss 0.0014 dev loss 0.7406 dev tag acc 98.64% dev head acc 96.76% dev deprel acc 97.67% | |
| [hops] 2024-09-23 23:36:01.771 | INFO | Epoch 61: train loss 0.0014 dev loss 0.7765 dev tag acc 98.63% dev head acc 96.75% dev deprel acc 97.65% | |
| [hops] 2024-09-23 23:37:32.476 | INFO | Epoch 62: train loss 0.0012 dev loss 0.7706 dev tag acc 98.64% dev head acc 96.73% dev deprel acc 97.67% | |
| [hops] 2024-09-23 23:38:59.167 | INFO | Epoch 63: train loss 0.0012 dev loss 0.7670 dev tag acc 98.64% dev head acc 96.74% dev deprel acc 97.67% | |
| [hops] 2024-09-23 23:39:04.307 | WARNING | You're using a RobertaTokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. | |
| [hops] 2024-09-23 23:39:12.221 | WARNING | You're using a RobertaTokenizerFast tokenizer. Please note that with a fast tokenizer, using the `__call__` method is faster than using a method to encode the text followed by a call to the `pad` method to get a padded encoding. | |
| [hops] 2024-09-23 23:39:13.462 | INFO | Metrics for GSD-camembertv2_base_p2_17k_last_layer+rand_seed=42 | |
| ─────────────────────────────── | |
| Split UPOS UAS LAS | |
| ─────────────────────────────── | |
| Dev 98.65 96.81 95.66 | |
| Test 98.66 95.77 94.32 | |
| ─────────────────────────────── | |