huggingartists

README.md

@@ -45,15 +45,15 @@ from datasets import load_dataset
 dataset = load_dataset("huggingartists/egor-letov")
 ```
 
-[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/hya2wr54/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
+[Explore the data](https://wandb.ai/huggingartists/huggingartists/runs/1turt8ph/artifacts), which is tracked with [W&B artifacts](https://docs.wandb.com/artifacts) at every step of the pipeline.
 
 ## Training procedure
 
 The model is based on a pre-trained [GPT-2](https://huggingface.co/gpt2) which is fine-tuned on Егор Летов (Egor Letov)'s lyrics.
 
-Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/2xmn5btx) for full transparency and reproducibility.
+Hyperparameters and metrics are recorded in the [W&B training run](https://wandb.ai/huggingartists/huggingartists/runs/797vcvtd) for full transparency and reproducibility.
 
-At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/2xmn5btx/artifacts) is logged and versioned.
+At the end of training, [the final model](https://wandb.ai/huggingartists/huggingartists/runs/797vcvtd/artifacts) is logged and versioned.
 
 ## How to use
 

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "gpt2",
+  "_name_or_path": "egor-letov",
   "activation_function": "gelu_new",
   "architectures": [
     "GPT2LMHeadModel"
@@ -18,7 +18,9 @@
   "n_inner": null,
   "n_layer": 12,
   "n_positions": 1024,
+  "reorder_and_upcast_attn": false,
   "resid_pdrop": 0.1,
+  "scale_attn_by_inverse_layer_idx": false,
   "scale_attn_weights": true,
   "summary_activation": null,
   "summary_first_dropout": 0.1,
@@ -35,7 +37,7 @@
     }
   },
   "torch_dtype": "float32",
-  "transformers_version": "4.10.0",
+  "transformers_version": "4.16.2",
   "use_cache": true,
   "vocab_size": 50257
 }

evaluation.txt

@@ -1 +1 @@
-{"eval_loss": 1.5267475843429565, "eval_runtime": 9.2267, "eval_samples_per_second": 22.11, "eval_steps_per_second": 2.818, "epoch": 2.0}
+{"eval_loss": 1.577836275100708, "eval_runtime": 12.1566, "eval_samples_per_second": 20.647, "eval_steps_per_second": 2.632, "epoch": 2.0}

flax_model.msgpack

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:40700a43fcf1b02a14f4bb8e44242664ec821de7f71e62ce8dcb245285432e78
+oid sha256:40bb754f5a4e532e1094072560b37d102559f7da5ab2ceca89da9c32a84a7004
 size 497764120

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79391dfa439eff2558b48fe749e7fa10dff70ea8eb79ebbc7dfd787f3429bc0e
-size 995603825
+oid sha256:64e600161fae48afedb06bb4666511c0231bf86944de08c4a4d428ec8abcab03
+size 995604017

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0a5d02a6e7f229347bd45652792c9a042d3aa5454fd46ec59c8d727f20244b83
+oid sha256:c71055ec92f1e3ee0ac98e2e1b4a77daead3f0fcf9222865833bcc60ad599f2a
 size 510403817

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:88f8d817916e1bb4a7ecefdd9143d2a982a32f6858c347d52f3eb4de40945973
+oid sha256:3e6a63d522e295e27b8f1e5effa2220fd354b7dd5efe831096984edfac6701c0
 size 14503

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee09584c5906ca038a9c2d5521c942dd2bd4958f8a31439fa1b9425bbc512cef
+oid sha256:bc07e42f7b3eb4494cdbc9bcb8911780143ce3b5a9cba5c3f2005a37846c6c2b
 size 623

tokenizer_config.json

@@ -1 +1 @@
-{"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "gpt2", "tokenizer_class": "GPT2Tokenizer"}
+{"unk_token": "<|endoftext|>", "bos_token": "<|endoftext|>", "eos_token": "<|endoftext|>", "add_prefix_space": false, "model_max_length": 1024, "special_tokens_map_file": null, "name_or_path": "huggingartists/egor-letov", "tokenizer_class": "GPT2Tokenizer"}

trainer_state.json

@@ -1,8 +1,8 @@
 {
-  "best_metric": 1.5952510833740234,
-  "best_model_checkpoint": "output/egor-letov/checkpoint-162",
-  "epoch": 1.0,
-  "global_step": 162,
+  "best_metric": 1.577836275100708,
+  "best_model_checkpoint": "output/egor-letov/checkpoint-316",
+  "epoch": 2.0,
+  "global_step": 316,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -206,11 +206,205 @@
       "eval_samples_per_second": 22.286,
       "eval_steps_per_second": 2.872,
       "step": 162
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.633985999863723e-07,
+      "loss": 1.7473,
+      "step": 165
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.94348327322787e-06,
+      "loss": 1.676,
+      "step": 170
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.881847639061565e-06,
+      "loss": 1.6984,
+      "step": 175
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.459348994545111e-06,
+      "loss": 1.7465,
+      "step": 180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.650532712235251e-06,
+      "loss": 1.8561,
+      "step": 185
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3423883622425528e-05,
+      "loss": 1.7407,
+      "step": 190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7742137247474782e-05,
+      "loss": 1.7568,
+      "step": 195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.256264781457008e-05,
+      "loss": 1.6805,
+      "step": 200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.783780941254068e-05,
+      "loss": 1.6899,
+      "step": 205
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3515526133505226e-05,
+      "loss": 1.7279,
+      "step": 210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.953972655637492e-05,
+      "loss": 1.7098,
+      "step": 215
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.585091749132818e-05,
+      "loss": 1.7613,
+      "step": 220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.238677151664272e-05,
+      "loss": 1.635,
+      "step": 225
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.908274250554988e-05,
+      "loss": 1.7108,
+      "step": 230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.587270306435089e-05,
+      "loss": 1.6292,
+      "step": 235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.268959758664963e-05,
+      "loss": 1.8167,
+      "step": 240
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.946610447433497e-05,
+      "loss": 1.6766,
+      "step": 245
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.613530098541655e-05,
+      "loss": 1.5776,
+      "step": 250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263132414287828e-05,
+      "loss": 1.7031,
+      "step": 255
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.889002117761157e-05,
+      "loss": 1.6967,
+      "step": 260
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001048495830818497,
+      "loss": 1.7448,
+      "step": 265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000110451155016322,
+      "loss": 1.677,
+      "step": 270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011563941754292792,
+      "loss": 1.7431,
+      "step": 275
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012036313294285455,
+      "loss": 1.7339,
+      "step": 280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012457565122486245,
+      "loss": 1.7054,
+      "step": 285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012823537082655537,
+      "loss": 1.6155,
+      "step": 290
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00013130614945888676,
+      "loss": 1.7792,
+      "step": 295
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001337576610365259,
+      "loss": 1.8149,
+      "step": 300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001355656951691469,
+      "loss": 1.686,
+      "step": 305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00013671239625595536,
+      "loss": 1.6878,
+      "step": 310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001371864398222265,
+      "loss": 1.6786,
+      "step": 315
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.577836275100708,
+      "eval_runtime": 12.0961,
+      "eval_samples_per_second": 20.75,
+      "eval_steps_per_second": 2.645,
+      "step": 316
     }
   ],
-  "max_steps": 162,
-  "num_train_epochs": 1,
-  "total_flos": 168402714624000.0,
+  "max_steps": 316,
+  "num_train_epochs": 2,
+  "total_flos": 329097314304000.0,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a45ea3ad5812b14388d7d3516c24ee1c0d156a8bbb28d67551985e62537ea611
-size 2671
+oid sha256:e5a982dd4f44bc59417f7e482b5d0cf9f6714929b958f9dd7f250c26a4889aca
+size 3055