publish trace-repro-gru-ar-s7
Browse files
trace-repro-gru-ar-s7/best/config.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arch": "gru",
|
| 3 |
+
"input_dim": 768,
|
| 4 |
+
"hidden_dim": 768,
|
| 5 |
+
"output_dim": 768,
|
| 6 |
+
"num_layers": 2,
|
| 7 |
+
"num_speakers": 4,
|
| 8 |
+
"dropout": 0.1,
|
| 9 |
+
"attention_mode": "autoregressive",
|
| 10 |
+
"max_turns": 64
|
| 11 |
+
}
|
trace-repro-gru-ar-s7/best/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ed169f32d7ca7865b838285a5b57987b26081bec0739770cff05d4a4e2c7cfb
|
| 3 |
+
size 33096040
|
trace-repro-gru-ar-s7/config.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arch": "gru",
|
| 3 |
+
"input_dim": 768,
|
| 4 |
+
"hidden_dim": 768,
|
| 5 |
+
"output_dim": 768,
|
| 6 |
+
"num_layers": 2,
|
| 7 |
+
"num_speakers": 4,
|
| 8 |
+
"dropout": 0.1,
|
| 9 |
+
"attention_mode": "autoregressive",
|
| 10 |
+
"max_turns": 64
|
| 11 |
+
}
|
trace-repro-gru-ar-s7/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7ed169f32d7ca7865b838285a5b57987b26081bec0739770cff05d4a4e2c7cfb
|
| 3 |
+
size 33096040
|
trace-repro-gru-ar-s7/trainlog.jsonl
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 1, "train_loss": 5.76159, "val_loss": 5.64092, "epoch_sec": 98.9}
|
| 2 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 2, "train_loss": 3.93123, "val_loss": 5.30879, "epoch_sec": 74.8}
|
| 3 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 3, "train_loss": 3.6976, "val_loss": 5.18958, "epoch_sec": 74.9}
|
| 4 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 4, "train_loss": 3.58216, "val_loss": 5.11987, "epoch_sec": 74.5}
|
| 5 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 5, "train_loss": 3.5043, "val_loss": 5.07226, "epoch_sec": 73.9}
|
| 6 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 6, "train_loss": 3.44516, "val_loss": 5.03666, "epoch_sec": 74.1}
|
| 7 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 7, "train_loss": 3.39735, "val_loss": 5.00846, "epoch_sec": 74.2}
|
| 8 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 8, "train_loss": 3.35743, "val_loss": 4.9852, "epoch_sec": 74.2}
|
| 9 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 9, "train_loss": 3.32365, "val_loss": 4.96603, "epoch_sec": 74.1}
|
| 10 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 10, "train_loss": 3.29504, "val_loss": 4.95319, "epoch_sec": 74.1}
|
| 11 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 11, "train_loss": 3.27109, "val_loss": 4.94096, "epoch_sec": 74.0}
|
| 12 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 12, "train_loss": 3.25098, "val_loss": 4.9267, "epoch_sec": 74.2}
|
| 13 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 13, "train_loss": 3.23445, "val_loss": 4.91989, "epoch_sec": 74.4}
|
| 14 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 14, "train_loss": 3.2215, "val_loss": 4.91294, "epoch_sec": 74.2}
|
| 15 |
+
{"model": "gru", "mode": "ar", "seed": 7, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 15, "train_loss": 3.21205, "val_loss": 4.90922, "epoch_sec": 74.4}
|