Training in progress, step 32000, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cb1d7e6c9c0b0a239d02e15a2159556c5bd1fcf1e8847f331a9a6e433d20fa5a
+oid sha256:585c793fcb177f73ccea5d9e03f556c9367be9b24af17619c9336aa9784900a3
 size 132187888

last-checkpoint/ar_diffusion_info.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6e86b3a858274139b3be961d3af1a11f158dbed04011a2fb0d226fa427bebe93
+oid sha256:70a55c79abf48cb04d5f4a5872743d51e805cb5cf801861bd90f35f89319726c
 size 1800

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e7c2116295a25f8bb18ac5af56b3b66bcefc07893fdd1dace52696c12337661
+oid sha256:0f58de48eab208cdcbb773a7a3efa7a8cc66fe9d9df03bf3d9123033730c12d8
 size 264665786

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79c133c2a3f6f8e5d4540624b02ef0bb23de1d12e242c3f6a4a6fbfc3892c66b
+oid sha256:36fd221e5d40799da39ad4f0eb8cd2d24c660462984d46b022ea86be7e548d3b
 size 14244

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8bd5d9ea98bbf4c19a3f4d1081add700e97c16f34e816ad12ce83b81a590f6a
+oid sha256:7daafea84c2a1039aac52712a160f9291aa4d9b6933d9afeee1976e296f5fb3a
 size 988

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3cc290e356b72b5d6c7ea8116aa7addab4b6ab1041682e12bf6b315f56282c40
+oid sha256:2f288b947beb52c2ae256c7cb95b541005fd607b063ecb73ec691524e7e48bb2
 size 1064

last-checkpoint/trainer_state.json

@@ -2,9 +2,9 @@
   "best_global_step": 31000,
   "best_metric": 0.6043956279754639,
   "best_model_checkpoint": "./qwen3-4b-ar-diffusion-checkpoints/checkpoint-31000",
-  "epoch": 2.422890546881009,
+  "epoch": 2.4613491269902315,
   "eval_steps": 250,
-  "global_step": 31500,
+  "global_step": 32000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -5426,6 +5426,92 @@
       "eval_samples_per_second": 22.349,
       "eval_steps_per_second": 5.587,
       "step": 31500
+    },
+    {
+      "epoch": 2.4267364048919315,
+      "grad_norm": 1.0685057640075684,
+      "learning_rate": 3.878139365763707e-05,
+      "loss": 0.6121,
+      "step": 31550
+    },
+    {
+      "epoch": 2.4305822629028535,
+      "grad_norm": 0.9418405890464783,
+      "learning_rate": 3.8521673635820586e-05,
+      "loss": 0.6377,
+      "step": 31600
+    },
+    {
+      "epoch": 2.434428120913776,
+      "grad_norm": 1.20501708984375,
+      "learning_rate": 3.826714801444044e-05,
+      "loss": 0.6373,
+      "step": 31650
+    },
+    {
+      "epoch": 2.4382739789246983,
+      "grad_norm": 0.8628917336463928,
+      "learning_rate": 3.800742799262395e-05,
+      "loss": 0.5955,
+      "step": 31700
+    },
+    {
+      "epoch": 2.4421198369356203,
+      "grad_norm": 1.0336881875991821,
+      "learning_rate": 3.774770797080747e-05,
+      "loss": 0.6184,
+      "step": 31750
+    },
+    {
+      "epoch": 2.4421198369356203,
+      "eval_loss": 0.6398035287857056,
+      "eval_runtime": 21.3292,
+      "eval_samples_per_second": 23.442,
+      "eval_steps_per_second": 5.861,
+      "step": 31750
+    },
+    {
+      "epoch": 2.4459656949465427,
+      "grad_norm": 0.8555242419242859,
+      "learning_rate": 3.748798794899099e-05,
+      "loss": 0.6232,
+      "step": 31800
+    },
+    {
+      "epoch": 2.4498115529574647,
+      "grad_norm": 0.7610741853713989,
+      "learning_rate": 3.722826792717451e-05,
+      "loss": 0.6292,
+      "step": 31850
+    },
+    {
+      "epoch": 2.453657410968387,
+      "grad_norm": 0.8955392837524414,
+      "learning_rate": 3.696854790535802e-05,
+      "loss": 0.6514,
+      "step": 31900
+    },
+    {
+      "epoch": 2.457503268979309,
+      "grad_norm": 0.8403136730194092,
+      "learning_rate": 3.670882788354155e-05,
+      "loss": 0.6168,
+      "step": 31950
+    },
+    {
+      "epoch": 2.4613491269902315,
+      "grad_norm": 1.146596908569336,
+      "learning_rate": 3.644910786172506e-05,
+      "loss": 0.6265,
+      "step": 32000
+    },
+    {
+      "epoch": 2.4613491269902315,
+      "eval_loss": 0.6403785943984985,
+      "eval_runtime": 22.2091,
+      "eval_samples_per_second": 22.513,
+      "eval_steps_per_second": 5.628,
+      "step": 32000
     }
   ],
   "logging_steps": 50,