publish trace-repro-lite-bidi-s42
Browse files
trace-repro-lite-bidi-s42/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": "bidirectional",
|
| 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-lite-bidi-s42/best/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:271679ef6a3aa4d6d222fed308d176483186139df2e1ed5fdde90aebbadeaee0
|
| 3 |
+
size 179808064
|
trace-repro-lite-bidi-s42/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": "bidirectional",
|
| 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-lite-bidi-s42/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a519d3ca0426a99de37f0a013c83f62b42bb94f5aa0a7cfa97f00146609bfbbc
|
| 3 |
+
size 179808064
|
trace-repro-lite-bidi-s42/trainlog.jsonl
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 1, "train_loss": 3.37579, "val_loss": 3.98217, "epoch_sec": 150.9}
|
| 2 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 2, "train_loss": 2.40307, "val_loss": 3.41739, "epoch_sec": 125.5}
|
| 3 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 3, "train_loss": 1.89353, "val_loss": 3.16368, "epoch_sec": 121.8}
|
| 4 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 4, "train_loss": 1.53528, "val_loss": 3.0565, "epoch_sec": 120.2}
|
| 5 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 5, "train_loss": 1.25024, "val_loss": 3.03929, "epoch_sec": 121.4}
|
| 6 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 6, "train_loss": 1.02291, "val_loss": 3.05041, "epoch_sec": 120.4}
|
| 7 |
+
{"model": "lite", "mode": "bidi", "seed": 42, "lr": 0.0002, "canary": false, "init": "scratch", "epoch": 7, "train_loss": 0.84218, "val_loss": 3.05034, "epoch_sec": 120.6}
|