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

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19db6af7a57039f898e5296f41227000b1602d75e9a2b425a610d754249804e6
+size 179808064

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

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d18ec9714465bf79676a569d11bd57b0d02248072cbfb0fa0f3390faa1ef3cda
+size 179808064

trace-repro-lite-ar-s42-xlc/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/best", "epoch": 1, "train_loss": 3.89083, "val_loss": 4.92364, "epoch_sec": 131.7}
+{"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/best", "epoch": 2, "train_loss": 3.25338, "val_loss": 4.73127, "epoch_sec": 122.4}
+{"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/best", "epoch": 3, "train_loss": 2.91089, "val_loss": 4.63046, "epoch_sec": 123.6}
+{"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/best", "epoch": 4, "train_loss": 2.61048, "val_loss": 4.61497, "epoch_sec": 123.4}
+{"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/best", "epoch": 5, "train_loss": 2.34988, "val_loss": 4.61032, "epoch_sec": 124.0}
+{"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/best", "epoch": 6, "train_loss": 2.12809, "val_loss": 4.65612, "epoch_sec": 125.2}
+{"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/best", "epoch": 7, "train_loss": 1.94355, "val_loss": 4.66559, "epoch_sec": 124.4}