Training in progress, step 1000, checkpoint

last-checkpoint/adapter_config.json

@@ -33,12 +33,12 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "gate_proj",
-    "k_proj",
-    "o_proj",
-    "down_proj",
     "v_proj",
     "q_proj",
+    "gate_proj",
+    "down_proj",
+    "o_proj",
+    "k_proj",
     "up_proj"
   ],
   "target_parameters": null,

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:056b42131c364ac6d3ffdcf649a5c97906ef4f0d8b378519839e223cdd8ac91d
+oid sha256:f7f16ebf88a0d95056235f2d36fa0dde45d635dd19fe373f7221414a1374b7f8
 size 84972248

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21b04b38bd9193dccba2b26c7971c57e8612c2d88b3fe83ba627fe8b6bf98a41
+oid sha256:296ded756d0f414f7e445c370d0d925cf67f696f3ea8d13bca350f070ddf214c
 size 43434405

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fa4695b71941286f2b5c53635fbeb413ba790d86d240df638f320993845abe31
+oid sha256:a01d50a1db8acb4ced09f825addc1c2a57ecdfe1a6e741863efa9e5c63d21121
 size 14917

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69fba46c77118e21066090a5469435f96321e6406988c306b041caec1e98a4c8
+oid sha256:f787fa81c178b3e54f8156ac38c6f8fbdc1bce6b672c879fefb79c4d09a69a02
 size 14917

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d789993928e049b3b0b113d88443633a86c205feae04d9e432a85efd6874b32
+oid sha256:c0255635c2d75d2f3bb5934ec2e22bac7b97c15231ed8cbeda2a9e8b220185d9
 size 1465

last-checkpoint/trainer_state.json

@@ -1,10 +1,10 @@
 {
-  "best_global_step": 750,
-  "best_metric": 0.3200623393058777,
-  "best_model_checkpoint": "/kaggle/working/obsidian_critic_qwen35_t4x2_unsloth/runs/obsidian_critic_full_epoch/checkpoint-750",
-  "epoch": 0.3325757995676515,
+  "best_global_step": 1000,
+  "best_metric": 0.26276224851608276,
+  "best_model_checkpoint": "/kaggle/working/obsidian_critic_qwen35_t4x2_unsloth/runs/obsidian_critic_full_epoch/checkpoint-1000",
+  "epoch": 0.4434343994235353,
   "eval_steps": 125,
-  "global_step": 750,
+  "global_step": 1000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -278,6 +278,92 @@
       "tokens_per_second": 368.35941630029333,
       "tokens_per_step": 1787.9106666666667,
       "total_tokens_seen": 1340933
+    },
+    {
+      "epoch": 0.3547475195388282,
+      "grad_norm": 0.5721789598464966,
+      "last_batch_tokens": 124,
+      "learning_rate": 7.231630894432527e-05,
+      "loss": 0.29953609466552733,
+      "lr": 7.22538412484033e-05,
+      "step": 800,
+      "tokens_per_second": 65.97096831279634,
+      "tokens_per_step": 98.35625,
+      "total_tokens_seen": 78685
+    },
+    {
+      "epoch": 0.376919239510005,
+      "grad_norm": 0.4275953471660614,
+      "last_batch_tokens": 266,
+      "learning_rate": 6.914223011522581e-05,
+      "loss": 0.27611801147460935,
+      "lr": 6.907774584760349e-05,
+      "step": 850,
+      "tokens_per_second": 76.59339331072898,
+      "tokens_per_step": 183.97411764705882,
+      "total_tokens_seen": 156378
+    },
+    {
+      "epoch": 0.38800509949559336,
+      "eval_loss": 0.28222641348838806,
+      "eval_runtime": 113.424,
+      "eval_samples_per_second": 3.2,
+      "eval_steps_per_second": 1.605,
+      "last_batch_tokens": 172,
+      "lr": 6.745388997609773e-05,
+      "step": 875,
+      "tokens_per_second": 114.49594753151979,
+      "tokens_per_step": 258.8742857142857,
+      "total_tokens_seen": 226515
+    },
+    {
+      "epoch": 0.39909095948118173,
+      "grad_norm": 0.5093332529067993,
+      "last_batch_tokens": 209,
+      "learning_rate": 6.587497507323132e-05,
+      "loss": 0.26179553985595705,
+      "lr": 6.580878811582379e-05,
+      "step": 900,
+      "tokens_per_second": 82.29563477689274,
+      "tokens_per_step": 298.55555555555554,
+      "total_tokens_seen": 268700
+    },
+    {
+      "epoch": 0.4212626794523585,
+      "grad_norm": 0.3912750482559204,
+      "last_batch_tokens": 103,
+      "learning_rate": 6.253044742254792e-05,
+      "loss": 0.25117488861083986,
+      "lr": 6.246287994523805e-05,
+      "step": 950,
+      "tokens_per_second": 79.79549481684828,
+      "tokens_per_step": 366.02947368421053,
+      "total_tokens_seen": 347728
+    },
+    {
+      "epoch": 0.4434343994235353,
+      "grad_norm": 0.4664643406867981,
+      "last_batch_tokens": 203,
+      "learning_rate": 5.9124926897487534e-05,
+      "loss": 0.25925636291503906,
+      "lr": 5.9056307789940357e-05,
+      "step": 1000,
+      "tokens_per_second": 76.53280228762407,
+      "tokens_per_step": 422.387,
+      "total_tokens_seen": 422387
+    },
+    {
+      "epoch": 0.4434343994235353,
+      "eval_loss": 0.26276224851608276,
+      "eval_runtime": 95.1275,
+      "eval_samples_per_second": 3.816,
+      "eval_steps_per_second": 1.913,
+      "last_batch_tokens": 172,
+      "lr": 5.9056307789940357e-05,
+      "step": 1000,
+      "tokens_per_second": 337.8095566509732,
+      "tokens_per_step": 454.525,
+      "total_tokens_seen": 454525
     }
   ],
   "logging_steps": 50,
@@ -306,7 +392,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 5.483273203482624e+16,
+  "total_flos": 7.329721439433523e+16,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2642ae75d5711b6c3ab84cc92e85358bfb92d0164676c852ae2a64540ed39871
+oid sha256:c39bdfd581a6de355d794e4165fae9336ff85f4eeb9590da60305570e0139f34
 size 5841