publish trace-repro-mpnet-lite-ar-s123
Browse files
trace-repro-mpnet-lite-ar-s123/best/config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"input_dim": 768,
|
| 3 |
+
"hidden_dim": 768,
|
| 4 |
+
"output_dim": 768,
|
| 5 |
+
"num_layers": 6,
|
| 6 |
+
"num_heads": 8,
|
| 7 |
+
"dropout": 0.1,
|
| 8 |
+
"max_turns": 64,
|
| 9 |
+
"attention_mode": "autoregressive",
|
| 10 |
+
"use_speaker_embeddings": true,
|
| 11 |
+
"num_speakers": 4,
|
| 12 |
+
"layer_norm": true,
|
| 13 |
+
"ff_dim": 3072,
|
| 14 |
+
"activation": "gelu",
|
| 15 |
+
"output_residual": false,
|
| 16 |
+
"arch": "v2",
|
| 17 |
+
"head_mode": "tied_continuous",
|
| 18 |
+
"head_transform": true
|
| 19 |
+
}
|
trace-repro-mpnet-lite-ar-s123/best/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19aa3843c144444a8736bff4e80e137249ee6dda4dfac9c9fea8e5e18a18137d
|
| 3 |
+
size 179808064
|
trace-repro-mpnet-lite-ar-s123/config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"input_dim": 768,
|
| 3 |
+
"hidden_dim": 768,
|
| 4 |
+
"output_dim": 768,
|
| 5 |
+
"num_layers": 6,
|
| 6 |
+
"num_heads": 8,
|
| 7 |
+
"dropout": 0.1,
|
| 8 |
+
"max_turns": 64,
|
| 9 |
+
"attention_mode": "autoregressive",
|
| 10 |
+
"use_speaker_embeddings": true,
|
| 11 |
+
"num_speakers": 4,
|
| 12 |
+
"layer_norm": true,
|
| 13 |
+
"ff_dim": 3072,
|
| 14 |
+
"activation": "gelu",
|
| 15 |
+
"output_residual": false,
|
| 16 |
+
"arch": "v2",
|
| 17 |
+
"head_mode": "tied_continuous",
|
| 18 |
+
"head_transform": true
|
| 19 |
+
}
|
trace-repro-mpnet-lite-ar-s123/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5594f348079a0faebe4080109ac68681163a276558ee6f3857b1714ff4803792
|
| 3 |
+
size 179808064
|
trace-repro-mpnet-lite-ar-s123/trainlog.jsonl
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 1, "train_loss": 5.17706, "val_loss": 5.49219, "epoch_sec": 81.5}
|
| 2 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 2, "train_loss": 4.1553, "val_loss": 5.16186, "epoch_sec": 65.3}
|
| 3 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 3, "train_loss": 3.73207, "val_loss": 5.0052, "epoch_sec": 65.2}
|
| 4 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 4, "train_loss": 3.35474, "val_loss": 4.93733, "epoch_sec": 65.0}
|
| 5 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 5, "train_loss": 2.96459, "val_loss": 4.9351, "epoch_sec": 65.1}
|
| 6 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 6, "train_loss": 2.60163, "val_loss": 4.94081, "epoch_sec": 64.9}
|
| 7 |
+
{"model": "lite", "mode": "ar", "seed": 123, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 7, "train_loss": 2.28184, "val_loss": 4.95362, "epoch_sec": 64.9}
|