publish trace-repro-lite-ar-s42-xlc-scratch

trace-repro-lite-ar-s42-xlc-scratch/best/config.json

@@ -0,0 +1,19 @@
+{
+  "input_dim": 768,
+  "hidden_dim": 768,
+  "output_dim": 768,
+  "num_layers": 6,
+  "num_heads": 8,
+  "dropout": 0.1,
+  "max_turns": 64,
+  "attention_mode": "autoregressive",
+  "use_speaker_embeddings": true,
+  "num_speakers": 4,
+  "layer_norm": true,
+  "ff_dim": 3072,
+  "activation": "gelu",
+  "output_residual": false,
+  "arch": "v2",
+  "head_mode": "tied_continuous",
+  "head_transform": true
+}

trace-repro-lite-ar-s42-xlc-scratch/best/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c14f89eea9dfb1f8e85814712b9f8121a15bf392ca2ef553350dad94d271a7cb
+size 179808064

trace-repro-lite-ar-s42-xlc-scratch/config.json

@@ -0,0 +1,19 @@
+{
+  "input_dim": 768,
+  "hidden_dim": 768,
+  "output_dim": 768,
+  "num_layers": 6,
+  "num_heads": 8,
+  "dropout": 0.1,
+  "max_turns": 64,
+  "attention_mode": "autoregressive",
+  "use_speaker_embeddings": true,
+  "num_speakers": 4,
+  "layer_norm": true,
+  "ff_dim": 3072,
+  "activation": "gelu",
+  "output_residual": false,
+  "arch": "v2",
+  "head_mode": "tied_continuous",
+  "head_transform": true
+}

trace-repro-lite-ar-s42-xlc-scratch/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d2127f75396d60cbb72bc2940fb91efc9a353f60405ed50784815848db12557
+size 179808064

trace-repro-lite-ar-s42-xlc-scratch/trainlog.jsonl

@@ -0,0 +1,7 @@
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 1, "train_loss": 3.90529, "val_loss": 4.93637, "epoch_sec": 136.7}
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 2, "train_loss": 3.27677, "val_loss": 4.74328, "epoch_sec": 129.4}
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 3, "train_loss": 2.94457, "val_loss": 4.63786, "epoch_sec": 129.6}
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 4, "train_loss": 2.6537, "val_loss": 4.61236, "epoch_sec": 129.6}
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 5, "train_loss": 2.40161, "val_loss": 4.6058, "epoch_sec": 129.5}
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 6, "train_loss": 2.18248, "val_loss": 4.6352, "epoch_sec": 129.6}
+{"model": "lite", "mode": "ar", "seed": 42, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-repro-scr/best", "epoch": 7, "train_loss": 1.99531, "val_loss": 4.64747, "epoch_sec": 129.7}