logs

qwen2.5-7b_c4_1B/logs/driver.log

@@ -0,0 +1,7 @@
+[2026-11-06T04:11:23Z] START seed 43
+[2026-11-06T09:09:12Z] END seed 43 bits_rc=0 std_rc=0
+[2026-11-06T09:09:12Z] START seed 44
+[2026-11-06T14:13:57Z] END seed 44 bits_rc=0 std_rc=0
+[2026-11-06T14:13:57Z] START seed 42
+[2026-11-06T19:19:15Z] END seed 42 bits_rc=1 std_rc=0
+[2026-11-06T19:19:15Z] ALL_SEEDS_DONE

qwen2.5-7b_c4_1B/logs/eval_s42_bits.log

@@ -0,0 +1,18 @@
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-3051 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.5794 ppl=97.46
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-6102 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.3016 ppl=73.82
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-9153 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1404 ppl=62.83
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-12204 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.0806 ppl=59.18
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-15255 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.0106 ppl=55.18
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-18306 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9205 ppl=50.42
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-21357 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7969 ppl=44.56
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-24408 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7111 ppl=40.90
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-27459 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6595 ppl=38.84
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-30510 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6495 ppl=38.45
+RESULT arm=bits ckpt=/root/traj/s3_1B/s42_bits/checkpoint-30517 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6495 ppl=38.45
+Traceback (most recent call last):
+    raise HFValidationError(
+huggingface_hub.errors.HFValidationError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/root/traj/s3_1B/s42_bits/final'. Use `repo_type` argument if needed.
+Traceback (most recent call last):
+    raise OSError(f"{e}") from e
+OSError: Repo id must be in the form 'repo_name' or 'namespace/repo_name': '/root/traj/s3_1B/s42_bits/final'. Use `repo_type` argument if needed.
+EVAL_DONE seed=42 arm=bits

qwen2.5-7b_c4_1B/logs/eval_s42_standard.log

@@ -0,0 +1,13 @@
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-3051 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.6833 ppl=108.12
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-6102 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.3516 ppl=77.60
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-9153 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.2188 ppl=67.95
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-12204 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1192 ppl=61.51
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-15255 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.0435 ppl=57.03
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-18306 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9489 ppl=51.88
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-21357 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.8499 ppl=46.99
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-24408 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7672 ppl=43.26
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-27459 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7267 ppl=41.54
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-30510 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7190 ppl=41.22
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/checkpoint-30517 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7190 ppl=41.22
+RESULT arm=standard ckpt=/root/traj/s3_1B/s42_std/final enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7190 ppl=41.22
+EVAL_DONE seed=42 arm=standard

qwen2.5-7b_c4_1B/logs/eval_s43_bits.log

@@ -0,0 +1,13 @@
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-3051 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.5754 ppl=97.07
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-6102 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.2951 ppl=73.34
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-9153 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1445 ppl=63.09
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-12204 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.0525 ppl=57.54
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-15255 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9833 ppl=53.69
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-18306 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.8962 ppl=49.21
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-21357 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7994 ppl=44.68
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-24408 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7156 ppl=41.08
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-27459 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6666 ppl=39.12
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-30510 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6562 ppl=38.72
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/checkpoint-30517 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6562 ppl=38.71
+RESULT arm=bits ckpt=/root/traj/s3_1B/s43_bits/final enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6562 ppl=38.71
+EVAL_DONE seed=43 arm=bits

qwen2.5-7b_c4_1B/logs/eval_s43_standard.log

@@ -0,0 +1,13 @@
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-3051 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.7183 ppl=111.97
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-6102 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.3820 ppl=79.99
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-9153 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.2301 ppl=68.73
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-12204 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1149 ppl=61.25
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-15255 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.0254 ppl=56.00
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-18306 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9166 ppl=50.23
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-21357 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.8071 ppl=45.02
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-24408 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7318 ppl=41.75
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-27459 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6936 ppl=40.19
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-30510 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6864 ppl=39.90
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/checkpoint-30517 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6864 ppl=39.90
+RESULT arm=standard ckpt=/root/traj/s3_1B/s43_std/final enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6864 ppl=39.90
+EVAL_DONE seed=43 arm=standard

qwen2.5-7b_c4_1B/logs/eval_s44_bits.log

@@ -0,0 +1,13 @@
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-3051 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.5904 ppl=98.53
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-6102 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.2836 ppl=72.50
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-9153 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1615 ppl=64.17
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-12204 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1434 ppl=63.02
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-15255 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9833 ppl=53.69
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-18306 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9304 ppl=50.93
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-21357 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.8077 ppl=45.05
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-24408 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7198 ppl=41.26
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-27459 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6802 ppl=39.65
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-30510 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6686 ppl=39.20
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/checkpoint-30517 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6685 ppl=39.19
+RESULT arm=bits ckpt=/root/traj/s3_1B/s44_bits/final enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.6685 ppl=39.19
+EVAL_DONE seed=44 arm=bits

qwen2.5-7b_c4_1B/logs/eval_s44_standard.log

@@ -0,0 +1,13 @@
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-3051 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.7436 ppl=114.85
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-6102 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.4125 ppl=82.48
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-9153 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.3004 ppl=73.73
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-12204 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1941 ppl=66.30
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-15255 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=4.1051 ppl=60.65
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-18306 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.9967 ppl=54.42
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-21357 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.8834 ppl=48.59
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-24408 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.8016 ppl=44.77
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-27459 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7614 ppl=43.01
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-30510 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7536 ppl=42.67
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/checkpoint-30517 enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7536 ppl=42.67
+RESULT arm=standard ckpt=/root/traj/s3_1B/s44_std/final enc=mlp depth=2 W=512 nbits=256 tokens=2048000 eval_loss=3.7536 ppl=42.67
+EVAL_DONE seed=44 arm=standard