publish trace-repro-lite-ar-s7-xlc

trace-repro-lite-ar-s7-xlc/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-s7-xlc/best/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93d1609893c9e9f20f744b2edfb263e008af5e58d66b0e73cd59dd98865d5e55
+size 179808064

trace-repro-lite-ar-s7-xlc/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-s7-xlc/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91bd6c7b115528223facc724eeba297f7ea040745d1369c98b3e669d27231b36
+size 179808064

trace-repro-lite-ar-s7-xlc/trainlog.jsonl

@@ -0,0 +1,6 @@
+{"model": "lite", "mode": "ar", "seed": 7, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-seed7-repro/best", "epoch": 1, "train_loss": 3.8826, "val_loss": 4.92238, "epoch_sec": 145.5}
+{"model": "lite", "mode": "ar", "seed": 7, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-seed7-repro/best", "epoch": 2, "train_loss": 3.24493, "val_loss": 4.72615, "epoch_sec": 134.1}
+{"model": "lite", "mode": "ar", "seed": 7, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-seed7-repro/best", "epoch": 3, "train_loss": 2.90269, "val_loss": 4.62389, "epoch_sec": 133.8}
+{"model": "lite", "mode": "ar", "seed": 7, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-seed7-repro/best", "epoch": 4, "train_loss": 2.60322, "val_loss": 4.59615, "epoch_sec": 134.1}
+{"model": "lite", "mode": "ar", "seed": 7, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-seed7-repro/best", "epoch": 5, "train_loss": 2.3439, "val_loss": 4.61091, "epoch_sec": 134.1}
+{"model": "lite", "mode": "ar", "seed": 7, "lr": 0.0001, "canary": false, "init": "/home/jfernandez/trace/models/contextual-turn-encoder-base-lite-ar-xlc-tm-seed7-repro/best", "epoch": 6, "train_loss": 2.12403, "val_loss": 4.63965, "epoch_sec": 133.4}