diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..975098f --- /dev/null +++ b/.gitignore @@ -0,0 +1,4 @@ +_pycache__/ +*.pt +!latest_checkpoint.pt +!latest_model.pt diff --git a/__pycache__/BucketManager.cpython-312.pyc b/__pycache__/BucketManager.cpython-312.pyc deleted file mode 100644 index 0a15216..0000000 Binary files a/__pycache__/BucketManager.cpython-312.pyc and /dev/null differ diff --git a/__pycache__/MyDataset.cpython-312.pyc b/__pycache__/MyDataset.cpython-312.pyc deleted file mode 100644 index 9c47c3b..0000000 Binary files a/__pycache__/MyDataset.cpython-312.pyc and /dev/null differ diff --git a/__pycache__/MyLSTM.cpython-312.pyc b/__pycache__/MyLSTM.cpython-312.pyc deleted file mode 100644 index 3d91993..0000000 Binary files a/__pycache__/MyLSTM.cpython-312.pyc and /dev/null differ diff --git a/__pycache__/MyLayer.cpython-312.pyc b/__pycache__/MyLayer.cpython-312.pyc deleted file mode 100644 index 48e2534..0000000 Binary files a/__pycache__/MyLayer.cpython-312.pyc and /dev/null differ diff --git a/__pycache__/MyTokenizer.cpython-312.pyc b/__pycache__/MyTokenizer.cpython-312.pyc deleted file mode 100644 index 5e4f5f9..0000000 Binary files a/__pycache__/MyTokenizer.cpython-312.pyc and /dev/null differ diff --git a/__pycache__/TryBatch.cpython-312.pyc b/__pycache__/TryBatch.cpython-312.pyc deleted file mode 100644 index dd76167..0000000 Binary files a/__pycache__/TryBatch.cpython-312.pyc and /dev/null differ diff --git a/model/checkpoints/checkpoint_000.pt b/model/checkpoints/checkpoint_000.pt deleted file mode 100644 index 90a4cf2..0000000 --- a/model/checkpoints/checkpoint_000.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:4a626e19c877d514a27eb83b25ff11d731abb42c2c16566eff84e3e5b2e66880 -size 654860808 diff --git a/model/checkpoints/checkpoint_001.pt b/model/checkpoints/checkpoint_001.pt deleted file mode 100644 index 31f477d..0000000 --- a/model/checkpoints/checkpoint_001.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:f8b739c8c5c0ab6751067b6bd4367d4fc0fcb727f72deed4b7dfac9dbdb63e7a -size 654860808 diff --git a/model/checkpoints/latest_checkpoint.pt b/model/checkpoints/latest_checkpoint.pt index c4f41d1..a4da440 100644 --- a/model/checkpoints/latest_checkpoint.pt +++ b/model/checkpoints/latest_checkpoint.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:7363ca8721702e734b921882d02c8341ab4c78cb6d39f50f107a9dee868a56cd +oid sha256:d1e78a4b661b300d2cc55bdc9c9e4665d7e32e8dd0457c20c8761d86c040830e size 654861162 diff --git a/model/checkpoints/latest_model.pt b/model/checkpoints/latest_model.pt index bd8cc4b..74441b1 100644 --- a/model/checkpoints/latest_model.pt +++ b/model/checkpoints/latest_model.pt @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:aca69df31f73d84a9a4a899c24326980cebaea9129b067e5965af665aa80a36e -size 304391906 +oid sha256:076451f3ef48ec505b33a40f58bc2d66ad9e37b60118e0587f3e6c99482c5b92 +size 411641634 diff --git a/model/checkpoints/model_000.pt b/model/checkpoints/model_000.pt deleted file mode 100644 index 5a178a8..0000000 --- a/model/checkpoints/model_000.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:47588dcd2fba5a7edd2b00c04acaf1afb8b86a179de25c8c876eb90a28c78609 -size 312698402 diff --git a/model/checkpoints/model_001.pt b/model/checkpoints/model_001.pt deleted file mode 100644 index 6658c43..0000000 --- a/model/checkpoints/model_001.pt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:5d66db070f6559a7abee27cafde233e8d2c1f02b839f280d50a917b5ff74fb22 -size 304391714 diff --git a/training.log b/training.log index 29f47c9..4d91601 100644 --- a/training.log +++ b/training.log @@ -782,3 +782,19 @@ 2025-06-27 01:05:26 | INFO | Epoch: 2 | Bucket: 131 | Loss: 6.208 | Acc: 17.589% 2025-06-27 01:05:37 | INFO | Validation | Loss: 5.938 | Acc: 20.194% 2025-06-27 01:05:37 | INFO | Processing bucket 1/131 src: (34, 38) tgt: (27, 30) shard: 1/1 +2025-06-27 02:40:24 | INFO | Processing bucket 1/131 src: (31, 34) tgt: (18, 21) shard: 1/1 +2025-06-27 02:40:37 | INFO | Epoch: 3 | Bucket: 1 | Loss: 6.326 | Acc: 14.815% +2025-06-27 02:40:46 | INFO | Validation | Loss: 5.942 | Acc: 20.102% +2025-06-27 02:40:46 | INFO | Processing bucket 2/131 src: (13, 16) tgt: (9, 12) shard: 2/2 +2025-06-27 02:41:24 | INFO | Epoch: 3 | Bucket: 2 | Loss: 6.219 | Acc: 18.670% +2025-06-27 02:41:34 | INFO | Validation | Loss: 6.026 | Acc: 18.729% +2025-06-27 02:41:34 | INFO | Processing bucket 3/131 src: (64, 73) tgt: (44, 49) shard: 1/1 +2025-06-27 02:42:24 | INFO | Epoch: 3 | Bucket: 3 | Loss: 6.038 | Acc: 16.838% +2025-06-27 02:42:34 | INFO | Validation | Loss: 5.949 | Acc: 19.959% +2025-06-27 02:42:34 | INFO | Processing bucket 4/131 src: (38, 42) tgt: (27, 30) shard: 1/1 +2025-06-27 02:43:30 | INFO | Epoch: 3 | Bucket: 4 | Loss: 6.134 | Acc: 16.568% +2025-06-27 02:43:38 | INFO | Validation | Loss: 5.937 | Acc: 20.255% +2025-06-27 02:43:38 | INFO | Processing bucket 5/131 src: (16, 19) tgt: (9, 12) shard: 1/1 +2025-06-27 02:44:05 | INFO | Epoch: 3 | Bucket: 5 | Loss: 6.213 | Acc: 19.143% +2025-06-27 02:44:15 | INFO | Validation | Loss: 5.977 | Acc: 19.437% +2025-06-27 02:44:15 | INFO | Processing bucket 6/131 src: (102, 129) tgt: (82, 99) shard: 1/1