publish trace-repro-gru-ar-s42
Browse files
trace-repro-gru-ar-s42/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-s42/best/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c241988c0b6f1c82ec30e84847453945741ff49bbad437edcad9b852fa92e4ce
|
| 3 |
+
size 33096040
|
trace-repro-gru-ar-s42/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-s42/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c241988c0b6f1c82ec30e84847453945741ff49bbad437edcad9b852fa92e4ce
|
| 3 |
+
size 33096040
|
trace-repro-gru-ar-s42/trainlog.jsonl
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 1, "train_loss": 5.77535, "val_loss": 5.65139, "epoch_sec": 91.1}
|
| 2 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 2, "train_loss": 3.93242, "val_loss": 5.30947, "epoch_sec": 73.4}
|
| 3 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 3, "train_loss": 3.69812, "val_loss": 5.19087, "epoch_sec": 73.4}
|
| 4 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 4, "train_loss": 3.58303, "val_loss": 5.12035, "epoch_sec": 73.4}
|
| 5 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 5, "train_loss": 3.50508, "val_loss": 5.07105, "epoch_sec": 73.3}
|
| 6 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 6, "train_loss": 3.44581, "val_loss": 5.0341, "epoch_sec": 73.4}
|
| 7 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 7, "train_loss": 3.39798, "val_loss": 5.00538, "epoch_sec": 73.6}
|
| 8 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 8, "train_loss": 3.3581, "val_loss": 4.98223, "epoch_sec": 73.2}
|
| 9 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 9, "train_loss": 3.32439, "val_loss": 4.96374, "epoch_sec": 73.3}
|
| 10 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 10, "train_loss": 3.29586, "val_loss": 4.94821, "epoch_sec": 73.4}
|
| 11 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 11, "train_loss": 3.27171, "val_loss": 4.93397, "epoch_sec": 73.4}
|
| 12 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 12, "train_loss": 3.25144, "val_loss": 4.92411, "epoch_sec": 73.4}
|
| 13 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 13, "train_loss": 3.23481, "val_loss": 4.91835, "epoch_sec": 73.2}
|
| 14 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 14, "train_loss": 3.22186, "val_loss": 4.9115, "epoch_sec": 72.8}
|
| 15 |
+
{"model": "gru", "mode": "ar", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 15, "train_loss": 3.21241, "val_loss": 4.90776, "epoch_sec": 72.7}
|