Upload codva2_think_reasoning_FINAL_step016410_config.json with huggingface_hub

codva2_think_reasoning_FINAL_step016410_config.json

@@ -0,0 +1,45 @@
+{
+  "vocab_size": 50304,
+  "d_model": 896,
+  "n_layers": 18,
+  "n_heads": 14,
+  "n_kv_heads": 2,
+  "max_len": 2048,
+  "rope_theta": 500000.0,
+  "window_size": 512,
+  "pattern_mult": 2.5,
+  "reason_mult": 0.75,
+  "reason_depth": 2,
+  "gate_hidden": 64,
+  "gate_init": 0.0,
+  "diff_lambda_init": 0.8,
+  "tie_embeddings": true,
+  "sft_lr": 3e-05,
+  "sft_min_lr": 3e-06,
+  "warmup": 250,
+  "epochs": 60,
+  "micro_batch": 2,
+  "grad_accum": 2,
+  "grad_clip": 1.0,
+  "wd": 0.01,
+  "log_every": 20,
+  "sample_every": 150,
+  "ckpt_every": 750,
+  "max_hours": 11.0,
+  "max_examples": 1096,
+  "end_token_weight": 15.0,
+  "tag_close_weight": 20.0,
+  "tag_open_weight": 15.0,
+  "min_thinking_chars": 80,
+  "min_reasoning_chars": 60,
+  "early_stop_window": 6,
+  "early_stop_open_pct": 80.0,
+  "early_stop_close_pct": 80.0,
+  "early_stop_end_pct": 80.0,
+  "early_stop_reason_pct": 70.0,
+  "early_stop_min_samples": 12,
+  "step": 16410,
+  "loss": 4.364421939849853,
+  "stop_reason": "max_steps_or_time",
+  "arch": "CodVa-2-Think-Reasoning-v3"
+}