Upload stage_3_mtetahwr/meta_001123.json with huggingface_hub

stage_3_mtetahwr/meta_001123.json

@@ -2,7 +2,7 @@
   "step": 1123,
   "val_bpb": 2.2258310665885443,
   "model_config": {
-    "run": "rva_6x256/stage_3",
+    "run": "rva_6x256/stage_3_mtetahwr",
     "wandb_group": "rva_6x256",
     "seed": 42,
     "device_type": "cuda",
@@ -62,7 +62,7 @@
     "warmup_ratio": 0.05,
     "warmdown_ratio": 0.4,
     "final_lr_frac": 0.0,
-    "resume_from_step": -1,
+    "resume_from_step": 1123,
     "eval_every": 250,
     "eval_tokens": 10485760,
     "core_metric_every": -1,
@@ -74,14 +74,15 @@
     "profile_step": 2,
     "profile_micro_step": 0,
     "memory_history_max_entries": 10000,
-    "model_tag": "rva_6x256/stage_3",
+    "model_tag": "rva_6x256/stage_3_mtetahwr",
+    "stage": 3,
     "n_layer": 6,
     "n_head": 16,
     "n_kv_head": 16,
     "n_embd": 256
   },
   "user_config": {
-    "run": "rva_6x256/stage_3",
+    "run": "rva_6x256/stage_3_mtetahwr",
     "wandb_group": "rva_6x256",
     "seed": 42,
     "device_type": "cuda",
@@ -141,7 +142,7 @@
     "warmup_ratio": 0.05,
     "warmdown_ratio": 0.4,
     "final_lr_frac": 0.0,
-    "resume_from_step": -1,
+    "resume_from_step": 1123,
     "eval_every": 250,
     "eval_tokens": 10485760,
     "core_metric_every": -1,
@@ -153,13 +154,15 @@
     "profile_step": 2,
     "profile_micro_step": 0,
     "memory_history_max_entries": 10000,
-    "model_tag": "rva_6x256/stage_3"
+    "model_tag": "rva_6x256/stage_3_mtetahwr",
+    "stage": 3
   },
+  "stage": 3,
   "device_batch_size": 8,
   "sequence_len": 1024,
   "dataloader_state_dict": {
     "pq_idx": 2,
-    "rg_idx": 16
+    "rg_idx": 20
   },
   "loop_state": {
     "min_val_bpb": 2.2258310665885443,