results: phase 7-19d CSVs and summaries
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- .gitattributes +1 -0
- results/csv/3seed_top3_results_20260310_2242.csv +10 -0
- results/csv/a10_replication_runs_20260322_0639.csv +28 -0
- results/csv/a_only_screen_runs_20260320_2335.csv +2 -0
- results/csv/all_variants_results_20260310_2002.csv +6 -0
- results/csv/chaos_confirmation_runs_20260321_2215.csv +13 -0
- results/csv/cheap_qa_confirmation_runs_20260320_1854.csv +5 -0
- results/csv/cheap_qa_followup_runs_20260320_0904.csv +4 -0
- results/csv/dot_gate_runs_20260324_0712.csv +7 -0
- results/csv/faithful_x_screen_runs_seed42_20260319_0733.csv +9 -0
- results/csv/faithful_x_screen_trajectories_seed42_20260319_0733.csv +329 -0
- results/csv/friday_night_runs_20260321_0104.csv +6 -0
- results/csv/fullq_followup_runs_20260318_0937.csv +5 -0
- results/csv/fullq_followup_summary_20260318_0937.csv +5 -0
- results/csv/fullq_followup_trajectories_20260318_0937.csv +165 -0
- results/csv/fullq_overnight_runs_20260317_2339.csv +3 -0
- results/csv/fullq_overnight_summary_20260317_2339.csv +3 -0
- results/csv/fullq_overnight_trajectories_20260317_2339.csv +83 -0
- results/csv/fullq_warmup3000_aggregate_summary_20260318_2145.csv +2 -0
- results/csv/fullq_warmup3000_remaining_runs_20260318_2145.csv +3 -0
- results/csv/fullq_warmup3000_remaining_trajectories_20260318_2145.csv +83 -0
- results/csv/fullq_warmup3000_seed42_runs_20260318_1839.csv +2 -0
- results/csv/fullq_warmup3000_seed42_summary_20260318_1839.csv +2 -0
- results/csv/fullq_warmup3000_seed42_trajectories_20260318_1839.csv +42 -0
- results/csv/isolation_missing_trajectories_20260313_1428.csv +7 -0
- results/csv/midtier_q_recovery_20260325_0738.csv +4 -0
- results/csv/midtierq_warmup3000_followup_runs_20260317_0943.csv +3 -0
- results/csv/midtierq_warmup3000_followup_summary_20260317_0943.csv +3 -0
- results/csv/midtierq_warmup3000_followup_trajectories_20260317_0943.csv +83 -0
- results/csv/midtierq_warmup_probe_runs_20260316_2327.csv +3 -0
- results/csv/midtierq_warmup_probe_summary_20260316_2327.csv +3 -0
- results/csv/midtierq_warmup_probe_trajectories_20260316_2327.csv +83 -0
- results/csv/openwebtext_5variants_10k_trajectories_v3_20260312_1551.csv +16 -0
- results/csv/openwebtext_trajectories_20260311_2057.csv +7 -0
- results/csv/p2_rerun_runs_20260323_1714.csv +10 -0
- results/csv/phase2_coded_completeness_runs_20260325_1113.csv +7 -0
- results/csv/phase3_bilinear_diag_runs_20260325_2226.csv +4 -0
- results/csv/phase4_chartfill_runs_20260326_0730.csv +10 -0
- results/csv/phase5_warmup500_toptiers_runs_20260326_1722.csv +16 -0
- results/csv/phase6a_grok_qafollowups_w2000_runs_20260327_1722.csv +7 -0
- results/csv/phase6b_grok_qafollowups_w500_runs_20260328_0028.csv +7 -0
- results/csv/priority1_runs_20260322_1355.csv +13 -0
- results/csv/priority2_runs_20260323_0623.csv +16 -0
- results/csv/qa_screen_results.csv +11 -0
- results/csv/qv_124m_3seed_results_20260310_0041.csv +1 -0
- results/csv/qv_124m_5000_results_20260310_0900.csv +3 -0
- results/csv/qv_3seed_results_20260310_0036.csv +7 -0
- results/csv/qv_long_results_20260309_2224.csv +3 -0
- results/csv/qv_openwebtext_results_20260311_0830.csv +3 -0
- results/csv/qv_results_20260309_2137.csv +5 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
results/figures/phase19b_qa_softqa_pd_w2000_6seeds_summary_20260403_0711_heatmaps.png filter=lfs diff=lfs merge=lfs -text
|
results/csv/3seed_top3_results_20260310_2242.csv
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Seed,Final_Val_Loss
|
| 2 |
+
none,1337,7.3218
|
| 3 |
+
none,42,7.3918
|
| 4 |
+
none,123,7.2759
|
| 5 |
+
vnorm,1337,6.4886
|
| 6 |
+
vnorm,42,6.6105
|
| 7 |
+
vnorm,123,6.5917
|
| 8 |
+
qvnorm,1337,6.4545
|
| 9 |
+
qvnorm,42,6.5394
|
| 10 |
+
qvnorm,123,6.3194
|
results/csv/a10_replication_runs_20260322_0639.csv
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,category,warmup,seed,status,hardware,tokens_per_iter,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
none,baseline,core,2000,42,ok,A10_24GB,4096,1658.66,123.59,4.4719,10000,4.4719,True,out-openwebtext_none_warmup2000_seed42_a100,
|
| 3 |
+
none,baseline,core,2000,1337,ok,A10_24GB,4096,1648.82,123.59,4.4576,10000,4.4576,True,out-openwebtext_none_warmup2000_seed1337_a100,
|
| 4 |
+
none,baseline,core,2000,123,ok,A10_24GB,4096,1664.42,123.59,4.4729,10000,4.4729,True,out-openwebtext_none_warmup2000_seed123_a100,
|
| 5 |
+
dynamic,cheap_q,core,2000,42,ok,A10_24GB,4096,1764.21,123.64,4.4168,10000,4.4168,True,out-openwebtext_dynamic_warmup2000_seed42_a100,
|
| 6 |
+
dynamic,cheap_q,core,2000,1337,ok,A10_24GB,4096,1768.59,123.64,4.4187,10000,4.4187,True,out-openwebtext_dynamic_warmup2000_seed1337_a100,
|
| 7 |
+
dynamic,cheap_q,core,2000,123,ok,A10_24GB,4096,1757.57,123.64,4.4312,9750,4.4484,False,out-openwebtext_dynamic_warmup2000_seed123_a100,
|
| 8 |
+
dynamic_qconditioned_mlp128,midtier_q,core,2000,42,ok,A10_24GB,4096,1813.42,123.78,4.4047,9750,4.4226,False,out-openwebtext_dynamic_qconditioned_mlp128_warmup2000_seed42_a100,
|
| 9 |
+
dynamic_qconditioned_mlp128,midtier_q,core,2000,1337,ok,A10_24GB,4096,1809.87,123.78,4.4111,9500,4.4173,False,out-openwebtext_dynamic_qconditioned_mlp128_warmup2000_seed1337_a100,
|
| 10 |
+
dynamic_qconditioned_mlp128,midtier_q,core,2000,123,ok,A10_24GB,4096,1811.18,123.78,4.3958,9750,4.4064,False,out-openwebtext_dynamic_qconditioned_mlp128_warmup2000_seed123_a100,
|
| 11 |
+
dynamic_qa_conditioned,cheap_qa,core,2000,42,ok,A10_24GB,4096,1774.11,123.69,4.4267,9500,4.4286,False,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed42_a100,
|
| 12 |
+
dynamic_qa_conditioned,cheap_qa,core,2000,1337,ok,A10_24GB,4096,1776.26,123.69,4.4256,10000,4.4256,True,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed1337_a100,
|
| 13 |
+
dynamic_qa_conditioned,cheap_qa,core,2000,123,ok,A10_24GB,4096,1778.2,123.69,4.393,9750,4.4087,False,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed123_a100,
|
| 14 |
+
dynamic_a_conditioned,a_only,core,2000,42,ok,A10_24GB,4096,1741.46,123.64,4.4374,10000,4.4374,True,out-openwebtext_dynamic_a_conditioned_warmup2000_seed42_a100,
|
| 15 |
+
dynamic_a_conditioned,a_only,core,2000,1337,ok,A10_24GB,4096,1744.44,123.64,4.4252,10000,4.4252,True,out-openwebtext_dynamic_a_conditioned_warmup2000_seed1337_a100,
|
| 16 |
+
dynamic_a_conditioned,a_only,core,2000,123,ok,A10_24GB,4096,1746.27,123.64,4.4404,9750,4.4554,False,out-openwebtext_dynamic_a_conditioned_warmup2000_seed123_a100,
|
| 17 |
+
vnorm,vnorm,fairness,2000,42,ok,A10_24GB,4096,1847.65,123.59,4.4934,10000,4.4934,True,out-openwebtext_vnorm_warmup2000_seed42_a100,
|
| 18 |
+
vnorm,vnorm,fairness,2000,1337,ok,A10_24GB,4096,1852.51,123.59,4.4847,10000,4.4847,True,out-openwebtext_vnorm_warmup2000_seed1337_a100,
|
| 19 |
+
vnorm,vnorm,fairness,2000,123,ok,A10_24GB,4096,1853.73,123.59,4.489,10000,4.489,True,out-openwebtext_vnorm_warmup2000_seed123_a100,
|
| 20 |
+
qvnorm,qvnorm,fairness,2000,42,ok,A10_24GB,4096,2061.64,123.59,4.4746,10000,4.4746,True,out-openwebtext_qvnorm_warmup2000_seed42_a100,
|
| 21 |
+
qvnorm,qvnorm,fairness,2000,1337,ok,A10_24GB,4096,2059.33,123.59,4.4672,10000,4.4672,True,out-openwebtext_qvnorm_warmup2000_seed1337_a100,
|
| 22 |
+
qvnorm,qvnorm,fairness,2000,123,ok,A10_24GB,4096,2060.6,123.59,4.4525,10000,4.4525,True,out-openwebtext_qvnorm_warmup2000_seed123_a100,
|
| 23 |
+
static_gate_prehead,static_prehead,fairness,2000,42,ok,A10_24GB,4096,1719.58,123.6,4.4569,10000,4.4569,True,out-openwebtext_static_gate_prehead_warmup2000_seed42_a100,
|
| 24 |
+
static_gate_prehead,static_prehead,fairness,2000,1337,ok,A10_24GB,4096,1719.8,123.6,4.4281,10000,4.4281,True,out-openwebtext_static_gate_prehead_warmup2000_seed1337_a100,
|
| 25 |
+
static_gate_prehead,static_prehead,fairness,2000,123,ok,A10_24GB,4096,1721.57,123.6,4.4464,10000,4.4464,True,out-openwebtext_static_gate_prehead_warmup2000_seed123_a100,
|
| 26 |
+
post_rmsnorm_y,post_rmsnorm,fairness,2000,42,ok,A10_24GB,4096,1855.27,123.59,4.8385,10000,4.8385,True,out-openwebtext_post_rmsnorm_y_warmup2000_seed42_a100,
|
| 27 |
+
post_rmsnorm_y,post_rmsnorm,fairness,2000,1337,ok,A10_24GB,4096,1850.92,123.59,4.8342,10000,4.8342,True,out-openwebtext_post_rmsnorm_y_warmup2000_seed1337_a100,
|
| 28 |
+
post_rmsnorm_y,post_rmsnorm,fairness,2000,123,ok,A10_24GB,4096,1851.69,123.59,4.8465,10000,4.8465,True,out-openwebtext_post_rmsnorm_y_warmup2000_seed123_a100,
|
results/csv/a_only_screen_runs_20260320_2335.csv
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,variant_short,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_a_conditioned,a_only,2000,42,ok,3749.96,123.64,4.4167,10000,4.4167,True,out-openwebtext_dynamic_a_conditioned_warmup2000_seed42,
|
results/csv/all_variants_results_20260310_2002.csv
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Final_Val_Loss,vs_Baseline
|
| 2 |
+
none,5.8126,+0.0000
|
| 3 |
+
vnorm,5.3412,-0.4714
|
| 4 |
+
qvnorm,5.3275,-0.4851
|
| 5 |
+
dynamic,5.9240,+0.1114
|
| 6 |
+
dynamic_swiglu,5.8899,+0.0773
|
results/csv/chaos_confirmation_runs_20260321_2215.csv
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
none,baseline,2000,1337,ok,3696.59,123.59,4.4618,9750,4.5041,False,out-openwebtext_none_warmup2000_seed1337_chaos,
|
| 3 |
+
none,baseline,2000,123,ok,3648.77,123.59,4.4813,9500,4.5018,False,out-openwebtext_none_warmup2000_seed123_chaos,
|
| 4 |
+
dynamic_ones_gate,ones_identity,2000,1337,ok,3585.71,123.59,4.4612,9750,4.5046,False,out-openwebtext_dynamic_ones_gate_warmup2000_seed1337_chaos,
|
| 5 |
+
dynamic_ones_gate,ones_identity,2000,123,ok,3556.0,123.59,4.4796,9500,4.5018,False,out-openwebtext_dynamic_ones_gate_warmup2000_seed123_chaos,
|
| 6 |
+
dynamic_a_conditioned,a_only,2000,1337,ok,3692.26,123.64,4.4399,9250,4.4415,False,out-openwebtext_dynamic_a_conditioned_warmup2000_seed1337_chaos,
|
| 7 |
+
dynamic_a_conditioned,a_only,2000,123,ok,3727.95,123.64,4.429,10000,4.429,True,out-openwebtext_dynamic_a_conditioned_warmup2000_seed123_chaos,
|
| 8 |
+
dynamic_random_gate,random_uniform,2000,1337,ok,3589.57,123.59,4.4948,9750,4.5373,False,out-openwebtext_dynamic_random_gate_warmup2000_seed1337_chaos,
|
| 9 |
+
dynamic_random_gate,random_uniform,2000,123,ok,3654.88,123.59,4.507,9500,4.5266,False,out-openwebtext_dynamic_random_gate_warmup2000_seed123_chaos,
|
| 10 |
+
dynamic_random_normal,random_normal,2000,1337,ok,3700.54,123.59,4.471,9750,4.5129,False,out-openwebtext_dynamic_random_normal_warmup2000_seed1337_chaos,
|
| 11 |
+
dynamic_random_normal,random_normal,2000,123,ok,3731.77,123.59,4.4824,9500,4.5023,False,out-openwebtext_dynamic_random_normal_warmup2000_seed123_chaos,
|
| 12 |
+
dynamic_bernoulli_gate,bernoulli_drop,2000,1337,ok,3734.81,123.59,4.65,9750,4.6947,False,out-openwebtext_dynamic_bernoulli_gate_warmup2000_seed1337_chaos,
|
| 13 |
+
dynamic_bernoulli_gate,bernoulli_drop,2000,123,ok,3731.55,123.59,4.6417,9500,4.6618,False,out-openwebtext_dynamic_bernoulli_gate_warmup2000_seed123_chaos,
|
results/csv/cheap_qa_confirmation_runs_20260320_1854.csv
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,variant_short,warmup,seed,note,status,runtime_sec,params_m,best_val,best_step,final_val,out_dir,error
|
| 2 |
+
dynamic_qa_conditioned,cheap_qa,3000,42,depth probe,ok,3758.93,123.69,4.3711,10000,4.3711,out-openwebtext_dynamic_qa_conditioned_warmup3000_seed42,
|
| 3 |
+
dynamic_qa_conditioned,cheap_qa,3000,1337,new,ok,3738.78,123.69,4.364,9750,4.4108,out-openwebtext_dynamic_qa_conditioned_warmup3000_seed1337,
|
| 4 |
+
dynamic_qa_conditioned,cheap_qa,3000,123,new,ok,3773.97,123.69,4.3944,9750,4.4029,out-openwebtext_dynamic_qa_conditioned_warmup3000_seed123,
|
| 5 |
+
dynamic_qa_conditioned,cheap_qa,2000,42,clean rerun replaces screen anchor,ok,3743.75,123.69,4.3906,10000,4.3906,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed42,
|
results/csv/cheap_qa_followup_runs_20260320_0904.csv
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,variant_short,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,out_dir,error
|
| 2 |
+
dynamic_qa_conditioned,cheap_qa,3000,42,ok,3738.68,123.69,4.3603,10000,4.3603,out-openwebtext_dynamic_qa_conditioned_warmup3000_seed42,
|
| 3 |
+
dynamic_qa_conditioned,cheap_qa,2000,1337,ok,3759.41,123.69,4.3824,9750,4.4278,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed1337,
|
| 4 |
+
dynamic_qa_conditioned,cheap_qa,2000,123,ok,3769.64,123.69,4.3914,9750,4.4034,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed123,
|
results/csv/dot_gate_runs_20260324_0712.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_dot_elementwise,dot_elementwise,2000,42,ok,3813.58,123.59,4.4574,9750,4.4707,False,out-openwebtext_dot_gate_shared,
|
| 3 |
+
dynamic_dot_elementwise,dot_elementwise,2000,1337,ok,3744.31,123.59,4.4374,9750,4.4806,False,out-openwebtext_dot_gate_shared,
|
| 4 |
+
dynamic_dot_elementwise,dot_elementwise,2000,123,ok,3736.11,123.59,4.4662,9500,4.4847,False,out-openwebtext_dot_gate_shared,
|
| 5 |
+
dynamic_dot_scalar,dot_scalar,2000,42,ok,3797.6,123.59,4.4646,9750,4.478,False,out-openwebtext_dot_gate_shared,
|
| 6 |
+
dynamic_dot_scalar,dot_scalar,2000,1337,ok,3794.8,123.59,4.4549,9750,4.4988,False,out-openwebtext_dot_gate_shared,
|
| 7 |
+
dynamic_dot_scalar,dot_scalar,2000,123,ok,3794.47,123.59,4.4608,9500,4.4814,False,out-openwebtext_dot_gate_shared,
|
results/csv/faithful_x_screen_runs_seed42_20260319_0733.csv
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,variant_short,warmup,seed,best_train,best_val,final_train,final_val,status,out_dir
|
| 2 |
+
dynamic_xconditioned_g1,global_x,500,42,4.9935,4.9626,4.9955,4.9626,ok,out-openwebtext_dynamic_xconditioned_g1_warmup500_seed42
|
| 3 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,4.8763,4.8963,4.8908,4.8981,ok,out-openwebtext_dynamic_x_g1_headspecific_elementwise_warmup500_seed42
|
| 4 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,4.9272,4.9085,4.9365,4.9722,ok,out-openwebtext_dynamic_x_g1_headspecific_headwise_warmup500_seed42
|
| 5 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,4.8222,4.8102,4.832,4.8102,ok,out-openwebtext_dynamic_x_g1_headshared_elementwise_warmup500_seed42
|
| 6 |
+
dynamic_xconditioned_g1,global_x,2000,42,4.5486,4.5339,4.5486,4.5339,ok,out-openwebtext_dynamic_xconditioned_g1_warmup2000_seed42
|
| 7 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,4.6015,4.6268,4.6119,4.6268,ok,out-openwebtext_dynamic_x_g1_headspecific_elementwise_warmup2000_seed42
|
| 8 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,4.5812,4.5605,4.5815,4.6138,ok,out-openwebtext_dynamic_x_g1_headspecific_headwise_warmup2000_seed42
|
| 9 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,4.5746,4.5698,4.5834,4.5698,ok,out-openwebtext_dynamic_x_g1_headshared_elementwise_warmup2000_seed42
|
results/csv/faithful_x_screen_trajectories_seed42_20260319_0733.csv
ADDED
|
@@ -0,0 +1,329 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,variant_short,warmup,seed,step,train_loss,val_loss
|
| 2 |
+
dynamic_xconditioned_g1,global_x,500,42,0,10.9285,10.9307
|
| 3 |
+
dynamic_xconditioned_g1,global_x,500,42,250,6.681,6.7057
|
| 4 |
+
dynamic_xconditioned_g1,global_x,500,42,500,6.4529,6.4631
|
| 5 |
+
dynamic_xconditioned_g1,global_x,500,42,750,6.2627,6.2601
|
| 6 |
+
dynamic_xconditioned_g1,global_x,500,42,1000,6.1161,6.1267
|
| 7 |
+
dynamic_xconditioned_g1,global_x,500,42,1250,6.0258,6.005
|
| 8 |
+
dynamic_xconditioned_g1,global_x,500,42,1500,5.9264,5.9946
|
| 9 |
+
dynamic_xconditioned_g1,global_x,500,42,1750,5.8838,5.9378
|
| 10 |
+
dynamic_xconditioned_g1,global_x,500,42,2000,5.8895,5.8633
|
| 11 |
+
dynamic_xconditioned_g1,global_x,500,42,2250,5.8071,5.8109
|
| 12 |
+
dynamic_xconditioned_g1,global_x,500,42,2500,5.7782,5.7703
|
| 13 |
+
dynamic_xconditioned_g1,global_x,500,42,2750,5.7698,5.7463
|
| 14 |
+
dynamic_xconditioned_g1,global_x,500,42,3000,5.6971,5.6975
|
| 15 |
+
dynamic_xconditioned_g1,global_x,500,42,3250,5.672,5.6764
|
| 16 |
+
dynamic_xconditioned_g1,global_x,500,42,3500,5.6089,5.6063
|
| 17 |
+
dynamic_xconditioned_g1,global_x,500,42,3750,5.5859,5.5788
|
| 18 |
+
dynamic_xconditioned_g1,global_x,500,42,4000,5.5528,5.5668
|
| 19 |
+
dynamic_xconditioned_g1,global_x,500,42,4250,5.4983,5.501
|
| 20 |
+
dynamic_xconditioned_g1,global_x,500,42,4500,5.4733,5.4612
|
| 21 |
+
dynamic_xconditioned_g1,global_x,500,42,4750,5.3997,5.4271
|
| 22 |
+
dynamic_xconditioned_g1,global_x,500,42,5000,5.4084,5.4
|
| 23 |
+
dynamic_xconditioned_g1,global_x,500,42,5250,5.3678,5.3535
|
| 24 |
+
dynamic_xconditioned_g1,global_x,500,42,5500,5.3313,5.3238
|
| 25 |
+
dynamic_xconditioned_g1,global_x,500,42,5750,5.2949,5.2749
|
| 26 |
+
dynamic_xconditioned_g1,global_x,500,42,6000,5.2752,5.2354
|
| 27 |
+
dynamic_xconditioned_g1,global_x,500,42,6250,5.2338,5.2216
|
| 28 |
+
dynamic_xconditioned_g1,global_x,500,42,6500,5.1951,5.2393
|
| 29 |
+
dynamic_xconditioned_g1,global_x,500,42,6750,5.1776,5.1795
|
| 30 |
+
dynamic_xconditioned_g1,global_x,500,42,7000,5.1525,5.1585
|
| 31 |
+
dynamic_xconditioned_g1,global_x,500,42,7250,5.1442,5.128
|
| 32 |
+
dynamic_xconditioned_g1,global_x,500,42,7500,5.1203,5.1089
|
| 33 |
+
dynamic_xconditioned_g1,global_x,500,42,7750,5.0916,5.1123
|
| 34 |
+
dynamic_xconditioned_g1,global_x,500,42,8000,5.0751,5.0459
|
| 35 |
+
dynamic_xconditioned_g1,global_x,500,42,8250,5.0847,5.0323
|
| 36 |
+
dynamic_xconditioned_g1,global_x,500,42,8500,5.0208,5.0198
|
| 37 |
+
dynamic_xconditioned_g1,global_x,500,42,8750,5.0394,5.0198
|
| 38 |
+
dynamic_xconditioned_g1,global_x,500,42,9000,5.0234,5.0173
|
| 39 |
+
dynamic_xconditioned_g1,global_x,500,42,9250,5.0079,4.9808
|
| 40 |
+
dynamic_xconditioned_g1,global_x,500,42,9500,4.9935,4.9875
|
| 41 |
+
dynamic_xconditioned_g1,global_x,500,42,9750,4.9936,4.9926
|
| 42 |
+
dynamic_xconditioned_g1,global_x,500,42,10000,4.9955,4.9626
|
| 43 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,0,10.9565,10.9575
|
| 44 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,250,6.7018,6.7042
|
| 45 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,500,6.4583,6.4619
|
| 46 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,750,6.2521,6.3139
|
| 47 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,1000,6.1761,6.1932
|
| 48 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,1250,6.0819,6.0604
|
| 49 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,1500,5.9793,6.0033
|
| 50 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,1750,5.8904,5.9108
|
| 51 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,2000,5.8694,5.8723
|
| 52 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,2250,5.8189,5.8029
|
| 53 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,2500,5.7536,5.7766
|
| 54 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,2750,5.7384,5.7311
|
| 55 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,3000,5.6557,5.657
|
| 56 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,3250,5.6297,5.6021
|
| 57 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,3500,5.5932,5.5654
|
| 58 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,3750,5.5342,5.5087
|
| 59 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,4000,5.4446,5.4706
|
| 60 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,4250,5.4053,5.4094
|
| 61 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,4500,5.3368,5.3625
|
| 62 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,4750,5.3792,5.3398
|
| 63 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,5000,5.2798,5.32
|
| 64 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,5250,5.2654,5.2574
|
| 65 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,5500,5.2478,5.2209
|
| 66 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,5750,5.1944,5.2115
|
| 67 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,6000,5.1667,5.1457
|
| 68 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,6250,5.1057,5.1203
|
| 69 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,6500,5.0884,5.1097
|
| 70 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,6750,5.0826,5.07
|
| 71 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,7000,5.0487,5.0325
|
| 72 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,7250,5.0676,4.9651
|
| 73 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,7500,5.0238,4.9861
|
| 74 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,7750,4.98,5.0162
|
| 75 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,8000,5.0003,4.9469
|
| 76 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,8250,4.9577,4.9679
|
| 77 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,8500,4.9145,4.9384
|
| 78 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,8750,4.9129,4.9056
|
| 79 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,9000,4.9252,4.9046
|
| 80 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,9250,4.9083,4.9065
|
| 81 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,9500,4.8763,4.9261
|
| 82 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,9750,4.91,4.8963
|
| 83 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,500,42,10000,4.8908,4.8981
|
| 84 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,0,10.9127,10.9097
|
| 85 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,250,6.6983,6.6957
|
| 86 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,500,6.4544,6.4295
|
| 87 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,750,6.2549,6.273
|
| 88 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,1000,6.1686,6.131
|
| 89 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,1250,6.0884,6.0431
|
| 90 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,1500,5.9795,5.9387
|
| 91 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,1750,5.9041,5.8966
|
| 92 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,2000,5.8965,5.8677
|
| 93 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,2250,5.8029,5.7975
|
| 94 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,2500,5.7392,5.7883
|
| 95 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,2750,5.7187,5.7354
|
| 96 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,3000,5.6972,5.7018
|
| 97 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,3250,5.6664,5.6483
|
| 98 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,3500,5.6253,5.6175
|
| 99 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,3750,5.5886,5.6211
|
| 100 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,4000,5.5561,5.5274
|
| 101 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,4250,5.5013,5.5361
|
| 102 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,4500,5.4481,5.4586
|
| 103 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,4750,5.4349,5.442
|
| 104 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,5000,5.4036,5.4031
|
| 105 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,5250,5.3798,5.3329
|
| 106 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,5500,5.3013,5.2927
|
| 107 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,5750,5.267,5.242
|
| 108 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,6000,5.2117,5.2015
|
| 109 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,6250,5.2358,5.2368
|
| 110 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,6500,5.1487,5.1937
|
| 111 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,6750,5.1526,5.1331
|
| 112 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,7000,5.1137,5.1211
|
| 113 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,7250,5.0976,5.0746
|
| 114 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,7500,5.0794,5.0563
|
| 115 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,7750,5.0402,5.0355
|
| 116 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,8000,5.0414,4.9917
|
| 117 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,8250,5.0001,4.985
|
| 118 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,8500,5.0275,4.996
|
| 119 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,8750,4.9722,4.9661
|
| 120 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,9000,5.0004,4.9775
|
| 121 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,9250,4.9272,4.9693
|
| 122 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,9500,5.0059,4.988
|
| 123 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,9750,4.9563,4.9085
|
| 124 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,500,42,10000,4.9365,4.9722
|
| 125 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,0,10.9641,10.9611
|
| 126 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,250,6.6946,6.7145
|
| 127 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,500,6.4063,6.4462
|
| 128 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,750,6.2785,6.2869
|
| 129 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,1000,6.1378,6.1604
|
| 130 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,1250,6.0541,6.0947
|
| 131 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,1500,5.9642,6.0087
|
| 132 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,1750,5.9412,5.9494
|
| 133 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,2000,5.8454,5.8485
|
| 134 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,2250,5.7756,5.7915
|
| 135 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,2500,5.733,5.7478
|
| 136 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,2750,5.7128,5.7005
|
| 137 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,3000,5.6187,5.6704
|
| 138 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,3250,5.6366,5.6538
|
| 139 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,3500,5.5494,5.5446
|
| 140 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,3750,5.4983,5.528
|
| 141 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,4000,5.4412,5.4151
|
| 142 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,4250,5.3906,5.3826
|
| 143 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,4500,5.3254,5.3364
|
| 144 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,4750,5.2991,5.3062
|
| 145 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,5000,5.2959,5.2653
|
| 146 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,5250,5.2017,5.241
|
| 147 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,5500,5.1919,5.1748
|
| 148 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,5750,5.1508,5.1648
|
| 149 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,6000,5.1504,5.1432
|
| 150 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,6250,5.1222,5.1119
|
| 151 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,6500,5.0735,5.0881
|
| 152 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,6750,5.0261,5.0535
|
| 153 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,7000,5.0331,5.0451
|
| 154 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,7250,4.9561,5.0177
|
| 155 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,7500,4.9575,4.9418
|
| 156 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,7750,4.9341,4.9648
|
| 157 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,8000,4.9301,4.9259
|
| 158 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,8250,4.8973,4.9217
|
| 159 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,8500,4.9018,4.9151
|
| 160 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,8750,4.88,4.914
|
| 161 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,9000,4.8645,4.8715
|
| 162 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,9250,4.8416,4.8834
|
| 163 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,9500,4.8879,4.8471
|
| 164 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,9750,4.8222,4.8594
|
| 165 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,500,42,10000,4.832,4.8102
|
| 166 |
+
dynamic_xconditioned_g1,global_x,2000,42,0,10.9285,10.9307
|
| 167 |
+
dynamic_xconditioned_g1,global_x,2000,42,250,6.9854,7.0066
|
| 168 |
+
dynamic_xconditioned_g1,global_x,2000,42,500,6.5715,6.5807
|
| 169 |
+
dynamic_xconditioned_g1,global_x,2000,42,750,6.3602,6.3526
|
| 170 |
+
dynamic_xconditioned_g1,global_x,2000,42,1000,6.2076,6.222
|
| 171 |
+
dynamic_xconditioned_g1,global_x,2000,42,1250,6.0832,6.0665
|
| 172 |
+
dynamic_xconditioned_g1,global_x,2000,42,1500,5.9804,6.0442
|
| 173 |
+
dynamic_xconditioned_g1,global_x,2000,42,1750,5.9,5.9566
|
| 174 |
+
dynamic_xconditioned_g1,global_x,2000,42,2000,5.8509,5.8208
|
| 175 |
+
dynamic_xconditioned_g1,global_x,2000,42,2250,5.7053,5.7133
|
| 176 |
+
dynamic_xconditioned_g1,global_x,2000,42,2500,5.6119,5.6063
|
| 177 |
+
dynamic_xconditioned_g1,global_x,2000,42,2750,5.5443,5.5253
|
| 178 |
+
dynamic_xconditioned_g1,global_x,2000,42,3000,5.4483,5.4465
|
| 179 |
+
dynamic_xconditioned_g1,global_x,2000,42,3250,5.4092,5.4155
|
| 180 |
+
dynamic_xconditioned_g1,global_x,2000,42,3500,5.3249,5.321
|
| 181 |
+
dynamic_xconditioned_g1,global_x,2000,42,3750,5.2971,5.2945
|
| 182 |
+
dynamic_xconditioned_g1,global_x,2000,42,4000,5.2519,5.2667
|
| 183 |
+
dynamic_xconditioned_g1,global_x,2000,42,4250,5.2004,5.2006
|
| 184 |
+
dynamic_xconditioned_g1,global_x,2000,42,4500,5.1653,5.1513
|
| 185 |
+
dynamic_xconditioned_g1,global_x,2000,42,4750,5.0773,5.1141
|
| 186 |
+
dynamic_xconditioned_g1,global_x,2000,42,5000,5.0762,5.0764
|
| 187 |
+
dynamic_xconditioned_g1,global_x,2000,42,5250,5.0349,5.0262
|
| 188 |
+
dynamic_xconditioned_g1,global_x,2000,42,5500,4.9893,4.99
|
| 189 |
+
dynamic_xconditioned_g1,global_x,2000,42,5750,4.9523,4.9388
|
| 190 |
+
dynamic_xconditioned_g1,global_x,2000,42,6000,4.9266,4.8984
|
| 191 |
+
dynamic_xconditioned_g1,global_x,2000,42,6250,4.8741,4.8675
|
| 192 |
+
dynamic_xconditioned_g1,global_x,2000,42,6500,4.8266,4.8669
|
| 193 |
+
dynamic_xconditioned_g1,global_x,2000,42,6750,4.7998,4.8028
|
| 194 |
+
dynamic_xconditioned_g1,global_x,2000,42,7000,4.7615,4.7785
|
| 195 |
+
dynamic_xconditioned_g1,global_x,2000,42,7250,4.7445,4.7401
|
| 196 |
+
dynamic_xconditioned_g1,global_x,2000,42,7500,4.7307,4.7219
|
| 197 |
+
dynamic_xconditioned_g1,global_x,2000,42,7750,4.6776,4.7009
|
| 198 |
+
dynamic_xconditioned_g1,global_x,2000,42,8000,4.6669,4.6356
|
| 199 |
+
dynamic_xconditioned_g1,global_x,2000,42,8250,4.661,4.6222
|
| 200 |
+
dynamic_xconditioned_g1,global_x,2000,42,8500,4.6093,4.6066
|
| 201 |
+
dynamic_xconditioned_g1,global_x,2000,42,8750,4.6142,4.5991
|
| 202 |
+
dynamic_xconditioned_g1,global_x,2000,42,9000,4.5913,4.5867
|
| 203 |
+
dynamic_xconditioned_g1,global_x,2000,42,9250,4.5846,4.5579
|
| 204 |
+
dynamic_xconditioned_g1,global_x,2000,42,9500,4.5591,4.5727
|
| 205 |
+
dynamic_xconditioned_g1,global_x,2000,42,9750,4.5635,4.5715
|
| 206 |
+
dynamic_xconditioned_g1,global_x,2000,42,10000,4.5486,4.5339
|
| 207 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,0,10.9565,10.9575
|
| 208 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,250,6.9816,6.9874
|
| 209 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,500,6.567,6.5727
|
| 210 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,750,6.339,6.3976
|
| 211 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,1000,6.2527,6.2724
|
| 212 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,1250,6.1274,6.1158
|
| 213 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,1500,6.0107,6.0441
|
| 214 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,1750,5.8995,5.9239
|
| 215 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,2000,5.8766,5.8786
|
| 216 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,2250,5.779,5.7616
|
| 217 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,2500,5.6469,5.6718
|
| 218 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,2750,5.6136,5.6036
|
| 219 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,3000,5.4992,5.5031
|
| 220 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,3250,5.4602,5.4335
|
| 221 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,3500,5.4257,5.3991
|
| 222 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,3750,5.3717,5.3444
|
| 223 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,4000,5.2787,5.3054
|
| 224 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,4250,5.2381,5.2438
|
| 225 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,4500,5.1564,5.1822
|
| 226 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,4750,5.1991,5.1632
|
| 227 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,5000,5.1017,5.1451
|
| 228 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,5250,5.092,5.0842
|
| 229 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,5500,5.0749,5.049
|
| 230 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,5750,5.0103,5.0291
|
| 231 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,6000,4.9769,4.9617
|
| 232 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,6250,4.9067,4.9273
|
| 233 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,6500,4.8786,4.9025
|
| 234 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,6750,4.8752,4.8605
|
| 235 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,7000,4.8269,4.8131
|
| 236 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,7250,4.8373,4.7423
|
| 237 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,7500,4.7856,4.7589
|
| 238 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,7750,4.7414,4.7772
|
| 239 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,8000,4.749,4.7025
|
| 240 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,8250,4.7067,4.7152
|
| 241 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,8500,4.6567,4.6792
|
| 242 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,8750,4.6532,4.6445
|
| 243 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,9000,4.6658,4.6499
|
| 244 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,9250,4.6286,4.6379
|
| 245 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,9500,4.6015,4.6553
|
| 246 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,9750,4.6376,4.6277
|
| 247 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elementwise,2000,42,10000,4.6119,4.6268
|
| 248 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,0,10.9127,10.9097
|
| 249 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,250,6.9873,6.9803
|
| 250 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,500,6.5718,6.5528
|
| 251 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,750,6.365,6.3872
|
| 252 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,1000,6.2593,6.2235
|
| 253 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,1250,6.1388,6.0973
|
| 254 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,1500,6.0082,5.9687
|
| 255 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,1750,5.9111,5.9075
|
| 256 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,2000,5.9014,5.8796
|
| 257 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,2250,5.7491,5.7435
|
| 258 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,2500,5.6446,5.696
|
| 259 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,2750,5.5724,5.5912
|
| 260 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,3000,5.5092,5.5137
|
| 261 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,3250,5.4552,5.4418
|
| 262 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,3500,5.3853,5.3771
|
| 263 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,3750,5.3334,5.3721
|
| 264 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,4000,5.3056,5.2778
|
| 265 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,4250,5.2317,5.2695
|
| 266 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,4500,5.1806,5.1952
|
| 267 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,4750,5.1592,5.1709
|
| 268 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,5000,5.1255,5.1269
|
| 269 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,5250,5.1116,5.0673
|
| 270 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,5500,5.0336,5.026
|
| 271 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,5750,4.9894,4.9661
|
| 272 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,6000,4.9273,4.9248
|
| 273 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,6250,4.9397,4.9513
|
| 274 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,6500,4.8519,4.8946
|
| 275 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,6750,4.8475,4.8334
|
| 276 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,7000,4.8032,4.813
|
| 277 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,7250,4.7836,4.7664
|
| 278 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,7500,4.7496,4.7445
|
| 279 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,7750,4.7069,4.7108
|
| 280 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,8000,4.6933,4.6771
|
| 281 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,8250,4.6735,4.6596
|
| 282 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,8500,4.6794,4.6504
|
| 283 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,8750,4.64,4.6217
|
| 284 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,9000,4.6491,4.6324
|
| 285 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,9250,4.5812,4.6298
|
| 286 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,9500,4.6427,4.6471
|
| 287 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,9750,4.6037,4.5605
|
| 288 |
+
dynamic_x_g1_headspecific_headwise,x_headspec_headwise,2000,42,10000,4.5815,4.6138
|
| 289 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,0,10.9641,10.9611
|
| 290 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,250,6.9937,7.0107
|
| 291 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,500,6.5326,6.5801
|
| 292 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,750,6.3834,6.3953
|
| 293 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,1000,6.2323,6.2582
|
| 294 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,1250,6.0958,6.1303
|
| 295 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,1500,5.9894,6.0375
|
| 296 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,1750,5.969,5.9695
|
| 297 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,2000,5.8489,5.8564
|
| 298 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,2250,5.7297,5.7429
|
| 299 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,2500,5.6311,5.6496
|
| 300 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,2750,5.5873,5.5705
|
| 301 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,3000,5.487,5.5337
|
| 302 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,3250,5.4648,5.489
|
| 303 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,3500,5.3934,5.3854
|
| 304 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,3750,5.3552,5.384
|
| 305 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,4000,5.2855,5.2549
|
| 306 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,4250,5.237,5.2203
|
| 307 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,4500,5.1687,5.1787
|
| 308 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,4750,5.1472,5.1547
|
| 309 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,5000,5.1402,5.1158
|
| 310 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,5250,5.0456,5.0839
|
| 311 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,5500,5.0281,5.0114
|
| 312 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,5750,4.9814,4.9981
|
| 313 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,6000,4.9769,4.9707
|
| 314 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,6250,4.9337,4.9203
|
| 315 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,6500,4.8868,4.9033
|
| 316 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,6750,4.8294,4.868
|
| 317 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,7000,4.8205,4.8353
|
| 318 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,7250,4.7408,4.8047
|
| 319 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,7500,4.7347,4.7252
|
| 320 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,7750,4.7083,4.7435
|
| 321 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,8000,4.6992,4.7044
|
| 322 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,8250,4.6594,4.6829
|
| 323 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,8500,4.6568,4.6685
|
| 324 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,8750,4.6396,4.6736
|
| 325 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,9000,4.6167,4.6269
|
| 326 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,9250,4.6025,4.6316
|
| 327 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,9500,4.6326,4.6013
|
| 328 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,9750,4.5746,4.6139
|
| 329 |
+
dynamic_x_g1_headshared_elementwise,x_headshared_elementwise,2000,42,10000,4.5834,4.5698
|
results/csv/friday_night_runs_20260321_0104.csv
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,description,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_a_conditioned,a_only,learned: A-only Linear(64->64),2000,42,ok,3665.81,123.64,4.4178,10000,4.4178,True,out-openwebtext_dynamic_a_conditioned_warmup2000_seed42,
|
| 3 |
+
dynamic_random_gate,random_uniform,"chaos: uniform random [0,1]",2000,42,ok,3576.11,123.59,4.5132,9750,4.5238,False,out-openwebtext_dynamic_random_gate_warmup2000_seed42,
|
| 4 |
+
dynamic_ones_gate,ones_identity,sanity: always 1.0 (= baseline),2000,42,ok,3530.33,123.59,4.4702,9750,4.4869,False,out-openwebtext_dynamic_ones_gate_warmup2000_seed42,
|
| 5 |
+
dynamic_random_normal,random_normal,"chaos: normal(0.5,0.2) clamped",2000,42,ok,3594.27,123.59,4.4822,9750,4.4982,False,out-openwebtext_dynamic_random_normal_warmup2000_seed42,
|
| 6 |
+
dynamic_bernoulli_gate,bernoulli_drop,spicy: binary 50% dropout gate,2000,42,ok,3626.68,123.59,4.6572,9750,4.6653,False,out-openwebtext_dynamic_bernoulli_gate_warmup2000_seed42,
|
results/csv/fullq_followup_runs_20260318_0937.csv
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,warmup,seed,best_val,final_val,status,out_dir
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,4.8264,4.8264,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup500_seed1337
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,4.8511,4.8606,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup500_seed123
|
| 4 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,4.9532,4.9532,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup200_seed1337
|
| 5 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,4.8501,4.8581,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup200_seed123
|
results/csv/fullq_followup_summary_20260318_0937.csv
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_short,warmup,seed,best_val,final_val,delta_vs_20pct_fullq,delta_vs_baseline,delta_vs_cheapq,delta_vs_midtierq,delta_vs_fullx,delta_vs_seed42
|
| 2 |
+
full_q,500,1337,4.8264,4.8264,0.2850999999999999,-0.024200000000000443,-0.011100000000000776,0.015099999999999447,-0.13570000000000082,-0.005000000000000782
|
| 3 |
+
full_q,500,123,4.8511,4.8606,0.31930000000000014,0.009999999999999787,0.023099999999999454,0.04929999999999968,-0.10150000000000059,0.02919999999999945
|
| 4 |
+
full_q,200,1337,4.9532,4.9532,0.41190000000000015,0.02289999999999992,0.04579999999999984,0.07160000000000011,-0.2012999999999998,0.11209999999999987
|
| 5 |
+
full_q,200,123,4.8501,4.8581,0.31680000000000064,-0.0721999999999996,-0.04929999999999968,-0.02349999999999941,-0.29639999999999933,0.017000000000000348
|
results/csv/fullq_followup_trajectories_20260318_0937.csv
ADDED
|
@@ -0,0 +1,165 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,warmup,seed,step,val_loss
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,0,10.9625
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,250,6.7345
|
| 4 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,500,6.4372
|
| 5 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,750,6.2124
|
| 6 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,1000,6.1096
|
| 7 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,1250,6.0303
|
| 8 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,1500,5.9489
|
| 9 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,1750,5.8628
|
| 10 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,2000,5.7928
|
| 11 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,2250,5.7469
|
| 12 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,2500,5.6964
|
| 13 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,2750,5.6629
|
| 14 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,3000,5.6319
|
| 15 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,3250,5.5479
|
| 16 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,3500,5.5196
|
| 17 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,3750,5.4359
|
| 18 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,4000,5.4325
|
| 19 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,4250,5.3398
|
| 20 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,4500,5.3601
|
| 21 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,4750,5.3053
|
| 22 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,5000,5.2247
|
| 23 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,5250,5.2288
|
| 24 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,5500,5.1832
|
| 25 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,5750,5.1615
|
| 26 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,6000,5.0919
|
| 27 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,6250,5.0708
|
| 28 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,6500,5.0584
|
| 29 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,6750,5.0351
|
| 30 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,7000,5.0018
|
| 31 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,7250,5.0026
|
| 32 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,7500,5.0114
|
| 33 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,7750,4.9566
|
| 34 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,8000,4.9493
|
| 35 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,8250,4.9055
|
| 36 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,8500,4.8896
|
| 37 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,8750,4.9149
|
| 38 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,9000,4.9008
|
| 39 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,9250,4.8527
|
| 40 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,9500,4.8503
|
| 41 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,9750,4.8621
|
| 42 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,10000,4.8264
|
| 43 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,0,10.9729
|
| 44 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,250,6.72
|
| 45 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,500,6.4281
|
| 46 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,750,6.2723
|
| 47 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,1000,6.1383
|
| 48 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,1250,5.9902
|
| 49 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,1500,5.9922
|
| 50 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,1750,5.8389
|
| 51 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,2000,5.8157
|
| 52 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,2250,5.7524
|
| 53 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,2500,5.7045
|
| 54 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,2750,5.6444
|
| 55 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,3000,5.6073
|
| 56 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,3250,5.563
|
| 57 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,3500,5.5243
|
| 58 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,3750,5.4685
|
| 59 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,4000,5.42
|
| 60 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,4250,5.3793
|
| 61 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,4500,5.3447
|
| 62 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,4750,5.2964
|
| 63 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,5000,5.2746
|
| 64 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,5250,5.2449
|
| 65 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,5500,5.1576
|
| 66 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,5750,5.1574
|
| 67 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,6000,5.1152
|
| 68 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,6250,5.1106
|
| 69 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,6500,5.0562
|
| 70 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,6750,5.0281
|
| 71 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,7000,5.0396
|
| 72 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,7250,4.9995
|
| 73 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,7500,5.0107
|
| 74 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,7750,4.9668
|
| 75 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,8000,4.9602
|
| 76 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,8250,4.9177
|
| 77 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,8500,4.923
|
| 78 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,8750,4.9212
|
| 79 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,9000,4.9001
|
| 80 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,9250,4.8768
|
| 81 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,9500,4.8511
|
| 82 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,9750,4.8703
|
| 83 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,10000,4.8606
|
| 84 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,0,10.9625
|
| 85 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,250,6.7087
|
| 86 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,500,6.3765
|
| 87 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,750,6.1892
|
| 88 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,1000,6.11
|
| 89 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,1250,6.035
|
| 90 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,1500,5.9607
|
| 91 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,1750,5.8811
|
| 92 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,2000,5.8273
|
| 93 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,2250,5.7938
|
| 94 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,2500,5.7679
|
| 95 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,2750,5.736
|
| 96 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,3000,5.7143
|
| 97 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,3250,5.6381
|
| 98 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,3500,5.6127
|
| 99 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,3750,5.5455
|
| 100 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,4000,5.5486
|
| 101 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,4250,5.4538
|
| 102 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,4500,5.4714
|
| 103 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,4750,5.4284
|
| 104 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,5000,5.344
|
| 105 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,5250,5.3523
|
| 106 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,5500,5.3049
|
| 107 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,5750,5.2804
|
| 108 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,6000,5.2129
|
| 109 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,6250,5.195
|
| 110 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,6500,5.1822
|
| 111 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,6750,5.1587
|
| 112 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,7000,5.1253
|
| 113 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,7250,5.1222
|
| 114 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,7500,5.136
|
| 115 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,7750,5.0795
|
| 116 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,8000,5.0739
|
| 117 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,8250,5.0254
|
| 118 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,8500,5.0181
|
| 119 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,8750,5.0396
|
| 120 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,9000,5.0243
|
| 121 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,9250,4.9779
|
| 122 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,9500,4.9768
|
| 123 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,9750,4.9864
|
| 124 |
+
dynamic_qconditioned_fullwidth,full_q,200,1337,10000,4.9532
|
| 125 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,0,10.9729
|
| 126 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,250,6.6686
|
| 127 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,500,6.3682
|
| 128 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,750,6.2508
|
| 129 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,1000,6.1312
|
| 130 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,1250,6.0023
|
| 131 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,1500,6.0061
|
| 132 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,1750,5.8611
|
| 133 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,2000,5.8466
|
| 134 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,2250,5.7777
|
| 135 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,2500,5.7237
|
| 136 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,2750,5.6581
|
| 137 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,3000,5.6304
|
| 138 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,3250,5.58
|
| 139 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,3500,5.5443
|
| 140 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,3750,5.493
|
| 141 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,4000,5.4316
|
| 142 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,4250,5.3915
|
| 143 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,4500,5.3482
|
| 144 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,4750,5.2972
|
| 145 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,5000,5.2705
|
| 146 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,5250,5.2437
|
| 147 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,5500,5.1602
|
| 148 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,5750,5.1566
|
| 149 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,6000,5.1155
|
| 150 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,6250,5.1084
|
| 151 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,6500,5.0533
|
| 152 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,6750,5.0256
|
| 153 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,7000,5.0355
|
| 154 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,7250,4.9936
|
| 155 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,7500,5.0082
|
| 156 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,7750,4.967
|
| 157 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,8000,4.9576
|
| 158 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,8250,4.916
|
| 159 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,8500,4.9188
|
| 160 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,8750,4.9187
|
| 161 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,9000,4.8965
|
| 162 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,9250,4.8731
|
| 163 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,9500,4.8501
|
| 164 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,9750,4.8668
|
| 165 |
+
dynamic_qconditioned_fullwidth,full_q,200,123,10000,4.8581
|
results/csv/fullq_overnight_runs_20260317_2339.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,best_val,final_val,status,out_dir
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,4.8314,4.8314,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup500_seed42
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,4.8411,4.8411,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup200_seed42
|
results/csv/fullq_overnight_summary_20260317_2339.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_short,seed,warmup,best_val,final_val,delta_vs_20pct_fullq,delta_vs_baseline,delta_vs_cheapq,delta_vs_midtierq,delta_vs_fullx
|
| 2 |
+
full_q,42,500,4.8314,4.8314,0.2901000000000007,-0.019199999999999662,-0.006099999999999994,0.02010000000000023,-0.13070000000000004
|
| 3 |
+
full_q,42,200,4.8411,4.8411,0.2998000000000003,-0.08919999999999995,-0.06630000000000003,-0.04049999999999976,-0.3133999999999997
|
results/csv/fullq_overnight_trajectories_20260317_2339.csv
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,step,val_loss
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,0,10.9291
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,250,6.7074
|
| 4 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,500,6.4463
|
| 5 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,750,6.2377
|
| 6 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,1000,6.1182
|
| 7 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,1250,6.0
|
| 8 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,1500,5.991
|
| 9 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,1750,5.9236
|
| 10 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,2000,5.8337
|
| 11 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,2250,5.7434
|
| 12 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,2500,5.6996
|
| 13 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,2750,5.645
|
| 14 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,3000,5.5911
|
| 15 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,3250,5.5609
|
| 16 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,3500,5.4858
|
| 17 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,3750,5.4586
|
| 18 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,4000,5.4324
|
| 19 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,4250,5.3734
|
| 20 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,4500,5.3244
|
| 21 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,4750,5.2905
|
| 22 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,5000,5.2597
|
| 23 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,5250,5.2156
|
| 24 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,5500,5.1874
|
| 25 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,5750,5.1412
|
| 26 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,6000,5.1055
|
| 27 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,6250,5.0865
|
| 28 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,6500,5.1017
|
| 29 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,6750,5.0482
|
| 30 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,7000,5.0257
|
| 31 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,7250,4.9959
|
| 32 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,7500,4.9787
|
| 33 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,7750,4.9788
|
| 34 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,8000,4.9163
|
| 35 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,8250,4.9012
|
| 36 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,8500,4.8893
|
| 37 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,8750,4.8891
|
| 38 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,9000,4.8878
|
| 39 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,9250,4.8501
|
| 40 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,9500,4.859
|
| 41 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,9750,4.8616
|
| 42 |
+
dynamic_qconditioned_fullwidth,full_q,42,500,10000,4.8314
|
| 43 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,0,10.9291
|
| 44 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,250,6.649
|
| 45 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,500,6.3899
|
| 46 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,750,6.205
|
| 47 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,1000,6.1151
|
| 48 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,1250,6.0016
|
| 49 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,1500,5.9994
|
| 50 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,1750,5.9404
|
| 51 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,2000,5.8532
|
| 52 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,2250,5.775
|
| 53 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,2500,5.711
|
| 54 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,2750,5.6551
|
| 55 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,3000,5.5955
|
| 56 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,3250,5.5693
|
| 57 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,3500,5.4883
|
| 58 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,3750,5.4559
|
| 59 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,4000,5.4372
|
| 60 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,4250,5.3676
|
| 61 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,4500,5.3216
|
| 62 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,4750,5.286
|
| 63 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,5000,5.2512
|
| 64 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,5250,5.2133
|
| 65 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,5500,5.1867
|
| 66 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,5750,5.1366
|
| 67 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,6000,5.1047
|
| 68 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,6250,5.085
|
| 69 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,6500,5.1024
|
| 70 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,6750,5.0511
|
| 71 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,7000,5.03
|
| 72 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,7250,5.0015
|
| 73 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,7500,4.9849
|
| 74 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,7750,4.9853
|
| 75 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,8000,4.9214
|
| 76 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,8250,4.9109
|
| 77 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,8500,4.8979
|
| 78 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,8750,4.8974
|
| 79 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,9000,4.8992
|
| 80 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,9250,4.8609
|
| 81 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,9500,4.8712
|
| 82 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,9750,4.8743
|
| 83 |
+
dynamic_qconditioned_fullwidth,full_q,42,200,10000,4.8411
|
results/csv/fullq_warmup3000_aggregate_summary_20260318_2145.csv
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_short,warmup,seeds,seed42,seed1337,seed123,mean_val,std_val,delta_vs_20pct_fullq,delta_vs_fullq_500,delta_vs_fullq_200,delta_vs_midtierq_3000,delta_vs_midtierq_2000,delta_vs_fullx_500,delta_vs_fullx_200
|
| 2 |
+
full_q,3000,"42,1337,123",4.5162,4.5052,4.5307,4.517366666666667,0.010442966798547056,-0.023933333333332918,-0.3221333333333334,-0.36673333333333336,-0.0031333333333334323,-0.023733333333333384,-0.44473333333333365,-0.6371333333333329
|
results/csv/fullq_warmup3000_remaining_runs_20260318_2145.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,best_val,final_val,status,out_dir
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,4.5052,4.5052,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup3000_seed1337
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,4.5307,4.5307,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup3000_seed123
|
results/csv/fullq_warmup3000_remaining_trajectories_20260318_2145.csv
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,step,val_loss
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,0,10.9625
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,250,7.1126
|
| 4 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,500,6.6561
|
| 5 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,750,6.4023
|
| 6 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,1000,6.283
|
| 7 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,1250,6.1413
|
| 8 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,1500,6.0186
|
| 9 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,1750,5.8857
|
| 10 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,2000,5.7886
|
| 11 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,2250,5.7168
|
| 12 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,2500,5.6441
|
| 13 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,2750,5.6023
|
| 14 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,3000,5.5725
|
| 15 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,3250,5.4885
|
| 16 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,3500,5.4267
|
| 17 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,3750,5.3412
|
| 18 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,4000,5.3207
|
| 19 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,4250,5.226
|
| 20 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,4500,5.2415
|
| 21 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,4750,5.1745
|
| 22 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,5000,5.0979
|
| 23 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,5250,5.0778
|
| 24 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,5500,5.0301
|
| 25 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,5750,4.9947
|
| 26 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,6000,4.9106
|
| 27 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,6250,4.8726
|
| 28 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,6500,4.8444
|
| 29 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,6750,4.8075
|
| 30 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,7000,4.7652
|
| 31 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,7250,4.7425
|
| 32 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,7500,4.7341
|
| 33 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,7750,4.6853
|
| 34 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,8000,4.6589
|
| 35 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,8250,4.6257
|
| 36 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,8500,4.6054
|
| 37 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,8750,4.5978
|
| 38 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,9000,4.5967
|
| 39 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,9250,4.5564
|
| 40 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,9500,4.5416
|
| 41 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,9750,4.5476
|
| 42 |
+
dynamic_qconditioned_fullwidth,full_q,1337,3000,10000,4.5052
|
| 43 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,0,10.9729
|
| 44 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,250,7.1368
|
| 45 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,500,6.66
|
| 46 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,750,6.4652
|
| 47 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,1000,6.2983
|
| 48 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,1250,6.1049
|
| 49 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,1500,6.0371
|
| 50 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,1750,5.8576
|
| 51 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,2000,5.8181
|
| 52 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,2250,5.7013
|
| 53 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,2500,5.6535
|
| 54 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,2750,5.579
|
| 55 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,3000,5.5445
|
| 56 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,3250,5.4812
|
| 57 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,3500,5.4434
|
| 58 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,3750,5.3689
|
| 59 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,4000,5.3144
|
| 60 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,4250,5.2717
|
| 61 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,4500,5.2181
|
| 62 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,4750,5.1507
|
| 63 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,5000,5.1308
|
| 64 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,5250,5.0959
|
| 65 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,5500,4.9903
|
| 66 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,5750,4.9849
|
| 67 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,6000,4.9189
|
| 68 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,6250,4.8963
|
| 69 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,6500,4.8236
|
| 70 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,6750,4.7828
|
| 71 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,7000,4.782
|
| 72 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,7250,4.7296
|
| 73 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,7500,4.7169
|
| 74 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,7750,4.6864
|
| 75 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,8000,4.6647
|
| 76 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,8250,4.6187
|
| 77 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,8500,4.6158
|
| 78 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,8750,4.6189
|
| 79 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,9000,4.5705
|
| 80 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,9250,4.5507
|
| 81 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,9500,4.5322
|
| 82 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,9750,4.541
|
| 83 |
+
dynamic_qconditioned_fullwidth,full_q,123,3000,10000,4.5307
|
results/csv/fullq_warmup3000_seed42_runs_20260318_1839.csv
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,best_val,final_val,status,out_dir
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,4.5162,4.5162,ok,out-openwebtext_dynamic_qconditioned_fullwidth_warmup3000_seed42
|
results/csv/fullq_warmup3000_seed42_summary_20260318_1839.csv
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_short,seed,warmup,best_val,final_val,delta_vs_20pct_fullq,delta_vs_fullq_500,delta_vs_fullq_200,delta_vs_midtierq_3000,delta_vs_midtierq_2000,delta_vs_fullx_500,delta_vs_fullx_200
|
| 2 |
+
full_q,42,3000,4.5162,4.5162,-0.025099999999999234,-0.3232999999999997,-0.36789999999999967,-0.0042999999999997485,-0.0248999999999997,-0.44589999999999996,-0.6382999999999992
|
results/csv/fullq_warmup3000_seed42_trajectories_20260318_1839.csv
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,step,val_loss
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,0,10.9291
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,250,7.1265
|
| 4 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,500,6.6704
|
| 5 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,750,6.4175
|
| 6 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,1000,6.2817
|
| 7 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,1250,6.0957
|
| 8 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,1500,6.0628
|
| 9 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,1750,5.9565
|
| 10 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,2000,5.8231
|
| 11 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,2250,5.7274
|
| 12 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,2500,5.646
|
| 13 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,2750,5.5951
|
| 14 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,3000,5.5313
|
| 15 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,3250,5.4967
|
| 16 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,3500,5.3947
|
| 17 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,3750,5.3626
|
| 18 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,4000,5.3374
|
| 19 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,4250,5.2645
|
| 20 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,4500,5.2081
|
| 21 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,4750,5.1678
|
| 22 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,5000,5.1226
|
| 23 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,5250,5.0678
|
| 24 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,5500,5.0232
|
| 25 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,5750,4.963
|
| 26 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,6000,4.9203
|
| 27 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,6250,4.8789
|
| 28 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,6500,4.8756
|
| 29 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,6750,4.8088
|
| 30 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,7000,4.782
|
| 31 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,7250,4.742
|
| 32 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,7500,4.7236
|
| 33 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,7750,4.6954
|
| 34 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,8000,4.6288
|
| 35 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,8250,4.613
|
| 36 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,8500,4.5969
|
| 37 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,8750,4.5848
|
| 38 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,9000,4.5738
|
| 39 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,9250,4.5426
|
| 40 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,9500,4.5571
|
| 41 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,9750,4.5526
|
| 42 |
+
dynamic_qconditioned_fullwidth,full_q,42,3000,10000,4.5162
|
results/csv/isolation_missing_trajectories_20260313_1428.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant_Seed,step_0,step_250,step_500,step_750,step_1000,step_1250,step_1500,step_1750,step_2000,step_2250,step_2500,step_2750,step_3000,step_3250,step_3500,step_3750,step_4000,step_4250,step_4500,step_4750,step_5000,step_5250,step_5500,step_5750,step_6000,step_6250,step_6500,step_6750,step_7000,step_7250,step_7500,step_7750,step_8000,step_8250,step_8500,step_8750,step_9000,step_9250,step_9500,step_9750,step_10000
|
| 2 |
+
post_rmsnorm_y_seed1337,10.9758,7.2382,6.7570,6.5159,6.3350,6.2654,6.1275,6.0888,6.0487,6.0079,5.9470,5.9118,5.8794,5.8075,5.7932,5.7474,5.7292,5.7031,5.6322,5.5924,5.5571,5.5075,5.4662,5.4216,5.3784,5.3923,5.3542,5.3136,5.2664,5.2455,5.1924,5.1888,5.1820,5.1593,5.1295,5.0968,5.1069,5.1082,5.0944,5.0565,5.0984
|
| 3 |
+
post_rmsnorm_y_seed42,10.9956,7.2506,6.7210,6.5283,6.3554,6.2420,6.1901,6.1218,6.0788,6.0025,5.9624,5.9135,5.8477,5.7949,5.8021,5.7340,5.6654,5.6051,5.5723,5.5300,5.4520,5.4300,5.4091,5.3524,5.3103,5.2883,5.2585,5.2002,5.2148,5.1539,5.1369,5.1198,5.1062,5.0692,5.1003,5.0454,5.0520,5.0098,4.9973,4.9974,5.0010
|
| 4 |
+
post_rmsnorm_y_seed123,10.8973,7.1877,6.7667,6.4901,6.3809,6.2586,6.1667,6.1304,6.0800,5.9984,5.9596,5.9131,5.8651,5.7830,5.7807,5.6989,5.6690,5.6271,5.5444,5.5194,5.4470,5.4299,5.4012,5.3721,5.3204,5.3285,5.2505,5.2405,5.1975,5.1781,5.1632,5.1189,5.0946,5.0712,5.0798,5.0554,5.0385,5.0224,5.0018,5.0341,5.0145
|
| 5 |
+
static_gate_seed1337,10.9790,7.0280,6.6008,6.3652,6.2015,6.1094,5.9543,5.9236,5.8373,5.7743,5.6601,5.5868,5.5347,5.4371,5.4033,5.3410,5.3115,5.2718,5.2033,5.1720,5.1362,5.0864,5.0475,4.9901,4.9563,4.9704,4.9365,4.8869,4.8400,4.8200,4.7597,4.7483,4.7450,4.7182,4.6917,4.6504,4.6644,4.6548,4.6430,4.6065,4.6490
|
| 6 |
+
static_gate_seed42,10.9677,7.0403,6.5675,6.3744,6.2316,6.0813,6.0089,5.9320,5.8813,5.7395,5.6568,5.5917,5.4968,5.4352,5.4312,5.3630,5.3010,5.2485,5.1957,5.1687,5.1039,5.0771,5.0564,5.0004,4.9615,4.9296,4.8995,4.8379,4.8402,4.7833,4.7667,4.7428,4.7223,4.6942,4.7173,4.6589,4.6718,4.6237,4.6168,4.6128,4.6215
|
| 7 |
+
static_gate_seed123,10.9585,6.9636,6.6099,6.3560,6.2473,6.0978,6.0054,5.9619,5.9047,5.7786,5.6883,5.6074,5.5604,5.4477,5.4461,5.3486,5.3220,5.2941,5.2018,5.1888,5.1048,5.0892,5.0641,5.0411,4.9756,4.9948,4.9209,4.9020,4.8494,4.8341,4.8064,4.7695,4.7404,4.7054,4.7166,4.6849,4.6692,4.6551,4.6237,4.6697,4.6431
|
results/csv/midtier_q_recovery_20260325_0738.csv
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
seed,status,best_val,best_step,final_val,stable,runtime_sec,error
|
| 2 |
+
42,ok,4.5733,10000,4.5733,True,~3600,
|
| 3 |
+
1337,ok,4.6151,10000,4.6151,True,3839.27,
|
| 4 |
+
123,ok,4.6143,10000,4.6143,True,3821.04,
|
results/csv/midtierq_warmup3000_followup_runs_20260317_0943.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,best_val,final_val,status,out_dir
|
| 2 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,4.5276,4.5276,ok,out-openwebtext_dynamic_qconditioned_mlp128_warmup3000_seed1337
|
| 3 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,4.5131,4.5131,ok,out-openwebtext_dynamic_qconditioned_mlp128_warmup3000_seed123
|
results/csv/midtierq_warmup3000_followup_summary_20260317_0943.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_short,seed,warmup,best_val,final_val,delta_vs_200_anchor,delta_vs_500_anchor,delta_vs_2000_anchor,delta_vs_3000_seed42
|
| 2 |
+
midtier_q,1337,3000,4.5276,4.5276,-0.3540000000000001,-0.2837000000000005,-0.013500000000000512,0.006799999999999251
|
| 3 |
+
midtier_q,123,3000,4.5131,4.5131,-0.36850000000000005,-0.29820000000000046,-0.02800000000000047,-0.007700000000000706
|
results/csv/midtierq_warmup3000_followup_trajectories_20260317_0943.csv
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,step,val_loss
|
| 2 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,0,10.983
|
| 3 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,250,7.1179
|
| 4 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,500,6.6854
|
| 5 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,750,6.4153
|
| 6 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,1000,6.2912
|
| 7 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,1250,6.1829
|
| 8 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,1500,6.0514
|
| 9 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,1750,5.9405
|
| 10 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,2000,5.8013
|
| 11 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,2250,5.7267
|
| 12 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,2500,5.6811
|
| 13 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,2750,5.5914
|
| 14 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,3000,5.5398
|
| 15 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,3250,5.5048
|
| 16 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,3500,5.3875
|
| 17 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,3750,5.3783
|
| 18 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,4000,5.3018
|
| 19 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,4250,5.2714
|
| 20 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,4500,5.2079
|
| 21 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,4750,5.1463
|
| 22 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,5000,5.1316
|
| 23 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,5250,5.085
|
| 24 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,5500,5.0114
|
| 25 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,5750,4.973
|
| 26 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,6000,4.9491
|
| 27 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,6250,4.8592
|
| 28 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,6500,4.8283
|
| 29 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,6750,4.8133
|
| 30 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,7000,4.7986
|
| 31 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,7250,4.7255
|
| 32 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,7500,4.693
|
| 33 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,7750,4.6413
|
| 34 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,8000,4.6347
|
| 35 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,8250,4.6247
|
| 36 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,8500,4.6204
|
| 37 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,8750,4.5707
|
| 38 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,9000,4.5727
|
| 39 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,9250,4.5399
|
| 40 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,9500,4.5459
|
| 41 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,9750,4.5386
|
| 42 |
+
dynamic_qconditioned_mlp128,midtier_q,1337,3000,10000,4.5276
|
| 43 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,0,11.0141
|
| 44 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,250,7.1221
|
| 45 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,500,6.6519
|
| 46 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,750,6.4652
|
| 47 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,1000,6.2818
|
| 48 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,1250,6.1601
|
| 49 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,1500,6.0735
|
| 50 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,1750,5.935
|
| 51 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,2000,5.8264
|
| 52 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,2250,5.7558
|
| 53 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,2500,5.6525
|
| 54 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,2750,5.6211
|
| 55 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,3000,5.5512
|
| 56 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,3250,5.4612
|
| 57 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,3500,5.4364
|
| 58 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,3750,5.354
|
| 59 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,4000,5.2767
|
| 60 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,4250,5.2503
|
| 61 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,4500,5.2171
|
| 62 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,4750,5.164
|
| 63 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,5000,5.1165
|
| 64 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,5250,5.1017
|
| 65 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,5500,4.9896
|
| 66 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,5750,4.9842
|
| 67 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,6000,4.9048
|
| 68 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,6250,4.8813
|
| 69 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,6500,4.8252
|
| 70 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,6750,4.7805
|
| 71 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,7000,4.7834
|
| 72 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,7250,4.7417
|
| 73 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,7500,4.6993
|
| 74 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,7750,4.6723
|
| 75 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,8000,4.6366
|
| 76 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,8250,4.6308
|
| 77 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,8500,4.6086
|
| 78 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,8750,4.5788
|
| 79 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,9000,4.5681
|
| 80 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,9250,4.572
|
| 81 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,9500,4.5582
|
| 82 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,9750,4.5369
|
| 83 |
+
dynamic_qconditioned_mlp128,midtier_q,123,3000,10000,4.5131
|
results/csv/midtierq_warmup_probe_runs_20260316_2327.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,best_val,final_val,status,out_dir
|
| 2 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,4.6618,4.6618,ok,out-openwebtext_dynamic_qconditioned_mlp128_warmup1000_seed42
|
| 3 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,4.5208,4.5208,ok,out-openwebtext_dynamic_qconditioned_mlp128_warmup3000_seed42
|
results/csv/midtierq_warmup_probe_summary_20260316_2327.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_short,seed,warmup,best_val,final_val,delta_vs_200_anchor,delta_vs_500_anchor,delta_vs_2000_anchor
|
| 2 |
+
midtier_q,42,1000,4.6618,4.6618,-0.21979999999999933,-0.14949999999999974,0.12070000000000025
|
| 3 |
+
midtier_q,42,3000,4.5208,4.5208,-0.36079999999999934,-0.29049999999999976,-0.020299999999999763
|
results/csv/midtierq_warmup_probe_trajectories_20260316_2327.csv
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant_long,variant_short,seed,warmup,step,val_loss
|
| 2 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,0,10.9324
|
| 3 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,250,6.8373
|
| 4 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,500,6.4826
|
| 5 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,750,6.3433
|
| 6 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,1000,6.2413
|
| 7 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,1250,6.0841
|
| 8 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,1500,5.9719
|
| 9 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,1750,5.8837
|
| 10 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,2000,5.8104
|
| 11 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,2250,5.7236
|
| 12 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,2500,5.6767
|
| 13 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,2750,5.5822
|
| 14 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,3000,5.4916
|
| 15 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,3250,5.448
|
| 16 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,3500,5.3643
|
| 17 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,3750,5.3256
|
| 18 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,4000,5.2584
|
| 19 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,4250,5.2488
|
| 20 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,4500,5.1826
|
| 21 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,4750,5.172
|
| 22 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,5000,5.151
|
| 23 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,5250,5.0783
|
| 24 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,5500,5.051
|
| 25 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,5750,4.9978
|
| 26 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,6000,4.9824
|
| 27 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,6250,4.9418
|
| 28 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,6500,4.9372
|
| 29 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,6750,4.8711
|
| 30 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,7000,4.8518
|
| 31 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,7250,4.8478
|
| 32 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,7500,4.784
|
| 33 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,7750,4.8038
|
| 34 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,8000,4.7793
|
| 35 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,8250,4.748
|
| 36 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,8500,4.7288
|
| 37 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,8750,4.7253
|
| 38 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,9000,4.7057
|
| 39 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,9250,4.6967
|
| 40 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,9500,4.6799
|
| 41 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,9750,4.6836
|
| 42 |
+
dynamic_qconditioned_mlp128,midtier_q,42,1000,10000,4.6618
|
| 43 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,0,10.9324
|
| 44 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,250,7.1289
|
| 45 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,500,6.6778
|
| 46 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,750,6.4488
|
| 47 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,1000,6.3278
|
| 48 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,1250,6.1341
|
| 49 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,1500,6.027
|
| 50 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,1750,5.9286
|
| 51 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,2000,5.8239
|
| 52 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,2250,5.7425
|
| 53 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,2500,5.6914
|
| 54 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,2750,5.5984
|
| 55 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,3000,5.5317
|
| 56 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,3250,5.4755
|
| 57 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,3500,5.3832
|
| 58 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,3750,5.3376
|
| 59 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,4000,5.2662
|
| 60 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,4250,5.2565
|
| 61 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,4500,5.1773
|
| 62 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,4750,5.1525
|
| 63 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,5000,5.1381
|
| 64 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,5250,5.0436
|
| 65 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,5500,5.0057
|
| 66 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,5750,4.9403
|
| 67 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,6000,4.9047
|
| 68 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,6250,4.8653
|
| 69 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,6500,4.8497
|
| 70 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,6750,4.7747
|
| 71 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,7000,4.7423
|
| 72 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,7250,4.7286
|
| 73 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,7500,4.6672
|
| 74 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,7750,4.6823
|
| 75 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,8000,4.6471
|
| 76 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,8250,4.6098
|
| 77 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,8500,4.5939
|
| 78 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,8750,4.5876
|
| 79 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,9000,4.5657
|
| 80 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,9250,4.5581
|
| 81 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,9500,4.5386
|
| 82 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,9750,4.5415
|
| 83 |
+
dynamic_qconditioned_mlp128,midtier_q,42,3000,10000,4.5208
|
results/csv/openwebtext_5variants_10k_trajectories_v3_20260312_1551.csv
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant_Seed,step_0,step_250,step_500,step_750,step_1000,step_1250,step_1500,step_1750,step_2000,step_2250,step_2500,step_2750,step_3000,step_3250,step_3500,step_3750,step_4000,step_4250,step_4500,step_4750,step_5000,step_5250,step_5500,step_5750,step_6000,step_6250,step_6500,step_6750,step_7000,step_7250,step_7500,step_7750,step_8000,step_8250,step_8500,step_8750,step_9000,step_9250,step_9500,step_9750,step_10000
|
| 2 |
+
none_seed1337,10.9985,7.0427,6.6101,6.3804,6.2195,6.1345,5.9773,5.9457,5.8792,5.8173,5.6914,5.6162,5.5619,5.4655,5.4417,5.3802,5.3517,5.3063,5.2470,5.2120,5.1748,5.1254,5.0858,5.0296,4.9934,5.0167,4.9756,4.9202,4.8768,4.8534,4.7929,4.7836,4.7763,4.7542,4.7252,4.6851,4.6986,4.6900,4.6775,4.6413,4.6842
|
| 3 |
+
none_seed42,10.9665,7.0584,6.5787,6.3861,6.2441,6.1054,6.0242,5.9616,5.8978,5.7755,5.6980,5.6264,5.5346,5.4711,5.4659,5.4022,5.3494,5.2878,5.2474,5.2158,5.1413,5.1166,5.1046,5.0440,5.0009,4.9756,4.9457,4.8847,4.8890,4.8308,4.8093,4.7855,4.7632,4.7374,4.7619,4.7021,4.7158,4.6666,4.6614,4.6538,4.6677
|
| 4 |
+
none_seed123,10.9279,6.9819,6.6188,6.3680,6.2589,6.1191,6.0234,5.9878,5.9272,5.8075,5.7195,5.6436,5.5814,5.4895,5.4749,5.3858,5.3672,5.3346,5.2436,5.2230,5.1440,5.1319,5.0951,5.0724,5.0130,5.0289,4.9560,4.9347,4.8856,4.8696,4.8424,4.8066,4.7771,4.7406,4.7512,4.7196,4.7033,4.6923,4.6579,4.7043,4.6784
|
| 5 |
+
vnorm_seed1337,11.0154,7.1036,6.6489,6.4211,6.2445,6.1793,6.0153,5.9784,5.9140,5.8349,5.7181,5.6412,5.5817,5.4850,5.4595,5.4050,5.3724,5.3240,5.2597,5.2283,5.1956,5.1460,5.1172,5.0524,5.0156,5.0410,5.0028,4.9553,4.9127,4.8945,4.8332,4.8276,4.8209,4.7938,4.7698,4.7287,4.7415,4.7351,4.7230,4.6863,4.7286
|
| 6 |
+
vnorm_seed42,10.9775,7.1299,6.6237,6.4207,6.2692,6.1379,6.0707,5.9960,5.9354,5.8183,5.7315,5.6576,5.5607,5.4927,5.4956,5.4160,5.3594,5.3005,5.2599,5.2275,5.1556,5.1321,5.1140,5.0640,5.0169,4.9933,4.9646,4.9086,4.9142,4.8561,4.8413,4.8167,4.7983,4.7707,4.7918,4.7351,4.7459,4.6992,4.6894,4.6854,4.6938
|
| 7 |
+
vnorm_seed123,10.8939,7.0354,6.6552,6.4028,6.2931,6.1743,6.0630,6.0251,5.9642,5.8455,5.7450,5.6771,5.5970,5.5033,5.4927,5.4039,5.3738,5.3403,5.2506,5.2287,5.1570,5.1459,5.1151,5.0868,5.0289,5.0451,4.9733,4.9571,4.9117,4.8932,4.8728,4.8297,4.8040,4.7690,4.7791,4.7489,4.7321,4.7190,4.6867,4.7323,4.7047
|
| 8 |
+
qvnorm_seed1337,11.0147,7.1093,6.6566,6.4222,6.2524,6.1781,6.0196,5.9674,5.8953,5.8081,5.6894,5.6169,5.5575,5.4608,5.4299,5.3745,5.3375,5.3028,5.2467,5.2097,5.1736,5.1267,5.0893,5.0345,4.9974,5.0170,4.9852,4.9355,4.8927,4.8764,4.8167,4.8122,4.8045,4.7803,4.7535,4.7131,4.7291,4.7228,4.7102,4.6733,4.7154
|
| 9 |
+
qvnorm_seed42,10.9763,7.1071,6.6305,6.4248,6.2721,6.1382,6.0595,5.9909,5.9210,5.8060,5.7071,5.6282,5.5315,5.4679,5.4656,5.3947,5.3340,5.2778,5.2350,5.2064,5.1385,5.1144,5.0973,5.0483,5.0038,4.9817,4.9557,4.8958,4.9019,4.8465,4.8306,4.8075,4.7903,4.7619,4.7830,4.7271,4.7398,4.6938,4.6866,4.6818,4.6918
|
| 10 |
+
qvnorm_seed123,10.8957,7.0406,6.6607,6.4137,6.3018,6.1765,6.0639,6.0045,5.9477,5.8278,5.7328,5.6514,5.5784,5.4782,5.4703,5.3758,5.3531,5.3227,5.2367,5.2144,5.1412,5.1319,5.0943,5.0775,5.0187,5.0343,4.9615,4.9494,4.9026,4.8844,4.8651,4.8205,4.7973,4.7651,4.7742,4.7450,4.7299,4.7161,4.6841,4.7305,4.7031
|
| 11 |
+
dynamic_seed1337,10.9212,6.9960,6.5602,6.3928,6.2228,6.1102,6.0117,5.8981,5.8084,5.6600,5.6061,5.5258,5.4382,5.4021,5.3230,5.2861,5.2420,5.2081,5.1674,5.1063,5.0717,5.0047,4.9737,4.9582,4.8944,4.8800,4.8479,4.8017,4.7083,4.7516,4.6941,4.6480,4.6518,4.6120,4.5834,4.6294,4.6206,4.5451,4.5591,4.5579,4.5483
|
| 12 |
+
dynamic_seed42,10.9612,7.0232,6.5924,6.4062,6.2685,6.1439,6.0359,5.9470,5.8034,5.6903,5.6027,5.5190,5.4688,5.4302,5.3246,5.3286,5.1969,5.1634,5.1234,5.1069,5.0542,5.0332,4.9572,4.9442,4.9185,4.8673,4.8501,4.8120,4.7834,4.7520,4.6760,4.6930,4.6518,4.6320,4.6219,4.6280,4.5809,4.5836,4.5561,4.5689,4.5213
|
| 13 |
+
dynamic_seed123,10.8906,7.0522,6.5668,6.4211,6.2459,6.1094,6.0207,5.8979,5.7960,5.7000,5.6304,5.5307,5.4322,5.3815,5.3474,5.3106,5.2396,5.1756,5.1245,5.0998,5.0455,5.0120,4.9984,4.9355,4.8924,4.8366,4.8475,4.8210,4.7818,4.7567,4.6909,4.6832,4.6477,4.6364,4.5870,4.6063,4.5910,4.5471,4.5479,4.5697,4.5383
|
| 14 |
+
dynamic_swiglu_seed1337,10.9439,6.9975,6.5536,6.3952,6.2409,6.1441,6.0786,5.9700,5.9040,5.7826,5.7253,5.6524,5.5621,5.5242,5.4519,5.4095,5.3653,5.3176,5.2729,5.2264,5.1827,5.1116,5.0825,5.0741,5.0149,5.0038,4.9748,4.9284,4.8393,4.8887,4.8192,4.7777,4.7738,4.7427,4.7118,4.7480,4.7413,4.6655,4.6715,4.6720,4.6629
|
| 15 |
+
dynamic_swiglu_seed42,10.9642,7.0132,6.5851,6.3997,6.2872,6.1806,6.0780,6.0012,5.8963,5.7963,5.7177,5.6363,5.6051,5.5600,5.4465,5.4331,5.3077,5.2802,5.2340,5.2059,5.1651,5.1367,5.0645,5.0576,5.0307,4.9893,4.9691,4.9254,4.9078,4.8792,4.7981,4.8144,4.7674,4.7491,4.7360,4.7357,4.6886,4.6960,4.6571,4.6676,4.6209
|
| 16 |
+
dynamic_swiglu_seed123,10.9377,7.0396,6.5705,6.4186,6.2644,6.1555,6.0699,5.9800,5.8751,5.8329,5.7793,5.6727,5.5825,5.5191,5.4775,5.4488,5.3814,5.3023,5.2522,5.2364,5.1768,5.1540,5.1255,5.0696,5.0291,4.9802,4.9904,4.9738,4.9400,4.9143,4.8521,4.8450,4.8066,4.7955,4.7354,4.7667,4.7430,4.6972,4.7012,4.7170,4.6845
|
results/csv/openwebtext_trajectories_20260311_2057.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant_Seed,step_0,step_250,step_500,step_750,step_1000,step_1250,step_1500,step_1750,step_2000,step_2250,step_2500,step_2750,step_3000,step_3250,step_3500,step_3750,step_4000,step_4250,step_4500,step_4750,step_5000
|
| 2 |
+
none_seed1337,10.9985,6.7059,6.4946,6.313,6.208,6.1625,6.0669,6.06,6.0251,5.998,5.9494,5.9373,5.9098,5.8606,5.8567,5.8405,5.8145,5.8281,5.786,5.7794,5.7645
|
| 3 |
+
none_seed42,10.9665,6.7184,6.4476,6.331,6.2137,6.1363,6.1127,6.0796,6.0503,6.0015,5.9757,5.9555,5.9188,5.8958,5.9229,5.8899,5.8496,5.8277,5.8147,5.8144,5.7681
|
| 4 |
+
none_seed123,10.9279,6.6556,6.4982,6.3082,6.2501,6.1633,6.1079,6.1143,6.0894,6.0417,6.0226,5.9996,5.9727,5.9242,5.9342,5.8925,5.9007,5.885,5.8525,5.8453,5.8073
|
| 5 |
+
qvnorm_seed1337,11.0147,6.7281,6.5016,6.3115,6.1836,6.1178,6.005,5.9823,5.906,5.8501,5.7639,5.7233,5.6702,5.592,5.5573,5.5184,5.4797,5.4679,5.4189,5.4031,5.3874
|
| 6 |
+
qvnorm_seed42,10.9763,6.7513,6.4533,6.3268,6.2005,6.1087,6.0574,5.9927,5.9407,5.8459,5.7725,5.7041,5.6293,5.5775,5.5723,5.5107,5.4561,5.4125,5.3904,5.3848,5.3315
|
| 7 |
+
qvnorm_seed123,10.8957,6.6741,6.5069,6.3107,6.2341,6.1215,6.0384,6.0155,5.9427,5.8586,5.8004,5.7285,5.6849,5.598,5.589,5.5121,5.5007,5.4703,5.4128,5.4085,5.3583
|
results/csv/p2_rerun_runs_20260323_1714.csv
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
static_gate_prehead,static_prehead,2000,42,ok,3670.55,123.6,4.455,9750,4.4699,False,out-openwebtext_rerun_shared,
|
| 3 |
+
static_gate_prehead,static_prehead,2000,1337,ok,3687.97,123.6,4.434,9750,4.4742,False,out-openwebtext_rerun_shared,
|
| 4 |
+
static_gate_prehead,static_prehead,2000,123,ok,3687.0,123.6,4.4559,9500,4.4777,False,out-openwebtext_rerun_shared,
|
| 5 |
+
dynamic_xconditioned_g1,full_x,2000,42,ok,3860.25,130.67,4.3897,10000,4.3897,True,out-openwebtext_rerun_shared,
|
| 6 |
+
dynamic_xconditioned_g1,full_x,2000,1337,ok,3873.53,130.67,4.3842,10000,4.3842,True,out-openwebtext_rerun_shared,
|
| 7 |
+
dynamic_xconditioned_g1,full_x,2000,123,failed,15.82,130.67,,,,False,out-openwebtext_rerun_shared,exit code 3221226505
|
| 8 |
+
dynamic_x_g1_headshared_elementwise,x_headshared,2000,42,ok,3733.58,123.64,4.4204,10000,4.4204,True,out-openwebtext_rerun_shared,
|
| 9 |
+
dynamic_x_g1_headshared_elementwise,x_headshared,2000,1337,ok,3728.6,123.64,4.4468,10000,4.4468,True,out-openwebtext_rerun_shared,
|
| 10 |
+
dynamic_x_g1_headshared_elementwise,x_headshared,2000,123,ok,3738.98,123.64,4.4367,10000,4.4367,True,out-openwebtext_rerun_shared,
|
results/csv/phase2_coded_completeness_runs_20260325_1113.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_qconditioned_fullwidth,full_q,2000,42,ok,3871.13,130.67,4.4027,10000,4.4027,True,out-openwebtext_phase2_full_q,
|
| 3 |
+
dynamic_qconditioned_fullwidth,full_q,2000,1337,ok,3791.18,130.67,4.4099,10000,4.4099,True,out-openwebtext_phase2_full_q,
|
| 4 |
+
dynamic_qconditioned_fullwidth,full_q,2000,123,ok,3786.03,130.67,4.4202,9500,4.4233,False,out-openwebtext_phase2_full_q,
|
| 5 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elem,2000,42,ok,3724.01,124.18,4.4638,10000,4.4638,True,out-openwebtext_phase2_x_headspec_elem,
|
| 6 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elem,2000,1337,ok,3718.84,124.18,4.4466,9750,4.4627,False,out-openwebtext_phase2_x_headspec_elem,
|
| 7 |
+
dynamic_x_g1_headspecific_elementwise,x_headspec_elem,2000,123,ok,3736.05,124.18,4.4247,9750,4.4393,False,out-openwebtext_phase2_x_headspec_elem,
|
results/csv/phase3_bilinear_diag_runs_20260325_2226.csv
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_qa_bilinear_diag,bilinear_diag,2000,42,ok,3883.85,123.59,4.4624,9750,4.4745,False,out-openwebtext_phase3_bilinear_diag,
|
| 3 |
+
dynamic_qa_bilinear_diag,bilinear_diag,2000,1337,ok,3837.97,123.59,4.4329,9750,4.4729,False,out-openwebtext_phase3_bilinear_diag,
|
| 4 |
+
dynamic_qa_bilinear_diag,bilinear_diag,2000,123,ok,3831.0,123.59,4.4596,9500,4.4824,False,out-openwebtext_phase3_bilinear_diag,
|
results/csv/phase4_chartfill_runs_20260326_0730.csv
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_qconditioned_fullwidth_headspecific,q_full_headspec,2000,42,ok,3811.02,130.67,4.4151,10000,4.4151,True,out-openwebtext_phase4_q_full_headspec,
|
| 3 |
+
dynamic_qconditioned_fullwidth_headspecific,q_full_headspec,2000,1337,ok,3797.53,130.67,4.4058,9750,4.4101,True,out-openwebtext_phase4_q_full_headspec,
|
| 4 |
+
dynamic_qconditioned_fullwidth_headspecific,q_full_headspec,2000,123,ok,3798.11,130.67,4.3964,9750,4.4366,True,out-openwebtext_phase4_q_full_headspec,
|
| 5 |
+
dynamic_xconditioned_fullwidth_headspecific,x_full_headspec,2000,42,ok,3798.84,130.67,4.4013,10000,4.4013,True,out-openwebtext_phase4_x_full_headspec,
|
| 6 |
+
dynamic_xconditioned_fullwidth_headspecific,x_full_headspec,2000,1337,ok,3801.74,130.67,4.4011,9750,4.4048,True,out-openwebtext_phase4_x_full_headspec,
|
| 7 |
+
dynamic_xconditioned_fullwidth_headspecific,x_full_headspec,2000,123,ok,3796.39,130.67,4.3867,9750,4.4239,True,out-openwebtext_phase4_x_full_headspec,
|
| 8 |
+
dynamic_qa_conditioned_headspecific,qa_headspec,2000,42,ok,3757.74,124.77,4.4004,9750,4.4005,True,out-openwebtext_phase4_qa_headspec,
|
| 9 |
+
dynamic_qa_conditioned_headspecific,qa_headspec,2000,1337,ok,3767.28,124.77,4.3945,9750,4.4295,True,out-openwebtext_phase4_qa_headspec,
|
| 10 |
+
dynamic_qa_conditioned_headspecific,qa_headspec,2000,123,ok,3771.44,124.77,4.3904,10000,4.3904,True,out-openwebtext_phase4_qa_headspec,
|
results/csv/phase5_warmup500_toptiers_runs_20260326_1722.csv
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_qa_conditioned_headspecific,qa_headspec,500,42,ok,3802.76,124.77,4.7306,10000,4.7306,False,out-openwebtext_phase5_w500_qa_headspec,
|
| 3 |
+
dynamic_qa_conditioned_headspecific,qa_headspec,500,1337,ok,3828.41,124.77,4.7612,9750,4.7981,False,out-openwebtext_phase5_w500_qa_headspec,
|
| 4 |
+
dynamic_qa_conditioned_headspecific,qa_headspec,500,123,ok,3830.43,124.77,4.6931,10000,4.6931,False,out-openwebtext_phase5_w500_qa_headspec,
|
| 5 |
+
dynamic_xconditioned_fullwidth_headspecific,x_full_headspec,500,42,ok,3899.73,130.67,4.8063,10000,4.8063,False,out-openwebtext_phase5_w500_x_full_headspec,
|
| 6 |
+
dynamic_xconditioned_fullwidth_headspecific,x_full_headspec,500,1337,ok,3871.37,130.67,4.8814,9750,4.9124,False,out-openwebtext_phase5_w500_x_full_headspec,
|
| 7 |
+
dynamic_xconditioned_fullwidth_headspecific,x_full_headspec,500,123,ok,3860.64,130.67,4.8183,9750,4.8416,False,out-openwebtext_phase5_w500_x_full_headspec,
|
| 8 |
+
dynamic_qa_conditioned_mlp128,mlp128_qa,500,42,ok,3902.14,123.88,4.6928,10000,4.6928,False,out-openwebtext_phase5_w500_mlp128_qa,
|
| 9 |
+
dynamic_qa_conditioned_mlp128,mlp128_qa,500,1337,ok,3905.2,123.88,4.6088,9750,4.6308,False,out-openwebtext_phase5_w500_mlp128_qa,
|
| 10 |
+
dynamic_qa_conditioned_mlp128,mlp128_qa,500,123,ok,3904.53,123.88,4.598,9750,4.6129,False,out-openwebtext_phase5_w500_mlp128_qa,
|
| 11 |
+
dynamic_qconditioned_fullwidth_headspecific,q_full_headspec,500,42,ok,3834.99,130.67,4.7184,10000,4.7184,False,out-openwebtext_phase5_w500_q_full_headspec,
|
| 12 |
+
dynamic_qconditioned_fullwidth_headspecific,q_full_headspec,500,1337,ok,3840.03,130.67,4.7927,9750,4.8213,False,out-openwebtext_phase5_w500_q_full_headspec,
|
| 13 |
+
dynamic_qconditioned_fullwidth_headspecific,q_full_headspec,500,123,ok,3839.74,130.67,4.6379,9750,4.6678,False,out-openwebtext_phase5_w500_q_full_headspec,
|
| 14 |
+
dynamic_qconditioned_fullwidth,full_q,500,42,ok,3853.37,130.67,4.725,10000,4.725,False,out-openwebtext_phase5_w500_full_q,
|
| 15 |
+
dynamic_qconditioned_fullwidth,full_q,500,1337,ok,3826.56,130.67,4.7695,10000,4.7695,False,out-openwebtext_phase5_w500_full_q,
|
| 16 |
+
dynamic_qconditioned_fullwidth,full_q,500,123,ok,3789.39,130.67,4.7423,9500,4.7474,False,out-openwebtext_phase5_w500_full_q,
|
results/csv/phase6a_grok_qafollowups_w2000_runs_20260327_1722.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_qa_conditioned_normed,qa_normed,2000,42,ok,4614.38,123.69,4.3882,10000,4.3882,False,out-openwebtext_phase6_w2000_qa_normed,
|
| 3 |
+
dynamic_qa_conditioned_normed,qa_normed,2000,1337,ok,4624.65,123.69,4.3881,9750,4.4334,False,out-openwebtext_phase6_w2000_qa_normed,
|
| 4 |
+
dynamic_qa_conditioned_normed,qa_normed,2000,123,ok,4569.71,123.69,4.4016,9750,4.4111,False,out-openwebtext_phase6_w2000_qa_normed,
|
| 5 |
+
dynamic_qa_conditioned_lowrank16,qa_lowrank,2000,42,ok,3849.29,123.62,4.411,9750,4.4223,False,out-openwebtext_phase6_w2000_qa_lowrank16,
|
| 6 |
+
dynamic_qa_conditioned_lowrank16,qa_lowrank,2000,1337,ok,3855.14,123.62,4.4024,9750,4.4145,False,out-openwebtext_phase6_w2000_qa_lowrank16,
|
| 7 |
+
dynamic_qa_conditioned_lowrank16,qa_lowrank,2000,123,ok,3832.51,123.62,4.4069,10000,4.4069,False,out-openwebtext_phase6_w2000_qa_lowrank16,
|
results/csv/phase6b_grok_qafollowups_w500_runs_20260328_0028.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
dynamic_qa_conditioned_normed,qa_normed,500,42,ok,4476.69,123.69,4.588,10000,4.588,False,out-openwebtext_phase6_w500_qa_normed,
|
| 3 |
+
dynamic_qa_conditioned_normed,qa_normed,500,1337,ok,4456.23,123.69,4.6118,9750,4.6508,False,out-openwebtext_phase6_w500_qa_normed,
|
| 4 |
+
dynamic_qa_conditioned_normed,qa_normed,500,123,ok,4460.28,123.69,4.5709,9750,4.5731,False,out-openwebtext_phase6_w500_qa_normed,
|
| 5 |
+
dynamic_qa_conditioned_lowrank16,qa_lowrank,500,42,ok,3813.82,123.62,4.6426,10000,4.6426,False,out-openwebtext_phase6_w500_qa_lowrank16,
|
| 6 |
+
dynamic_qa_conditioned_lowrank16,qa_lowrank,500,1337,ok,3837.27,123.62,4.6158,9750,4.6289,False,out-openwebtext_phase6_w500_qa_lowrank16,
|
| 7 |
+
dynamic_qa_conditioned_lowrank16,qa_lowrank,500,123,ok,3823.12,123.62,4.6282,10000,4.6282,False,out-openwebtext_phase6_w500_qa_lowrank16,
|
results/csv/priority1_runs_20260322_1355.csv
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
none,baseline,2000,42,ok,3635.68,123.59,4.4728,9750,4.4861,False,out-openwebtext_none_warmup2000_seed42_p1,
|
| 3 |
+
none,baseline,2000,1337,ok,3590.16,123.59,4.4609,9750,4.5048,False,out-openwebtext_none_warmup2000_seed1337_p1,
|
| 4 |
+
none,baseline,2000,123,ok,3590.42,123.59,4.4786,9500,4.501,False,out-openwebtext_none_warmup2000_seed123_p1,
|
| 5 |
+
dynamic_qconditioned_mlp128,midtier_q,2000,42,ok,3896.93,123.78,4.4085,10000,4.4085,True,out-openwebtext_dynamic_qconditioned_mlp128_warmup2000_seed42_p1,
|
| 6 |
+
dynamic_qconditioned_mlp128,midtier_q,2000,1337,ok,3887.93,123.78,4.4256,10000,4.4256,True,out-openwebtext_dynamic_qconditioned_mlp128_warmup2000_seed1337_p1,
|
| 7 |
+
dynamic_qconditioned_mlp128,midtier_q,2000,123,ok,3872.02,123.78,4.4065,10000,4.4065,True,out-openwebtext_dynamic_qconditioned_mlp128_warmup2000_seed123_p1,
|
| 8 |
+
dynamic_qa_conditioned,cheap_qa,2000,42,ok,3833.38,123.69,4.383,10000,4.383,True,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed42_p1,
|
| 9 |
+
dynamic_qa_conditioned,cheap_qa,2000,1337,ok,3813.97,123.69,4.3907,9750,4.4321,False,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed1337_p1,
|
| 10 |
+
dynamic_qa_conditioned,cheap_qa,2000,123,ok,3790.71,123.69,4.3953,9750,4.4047,False,out-openwebtext_dynamic_qa_conditioned_warmup2000_seed123_p1,
|
| 11 |
+
dynamic,cheap_q,2000,42,ok,3730.92,123.64,4.4069,10000,4.4069,True,out-openwebtext_dynamic_warmup2000_seed42_p1,
|
| 12 |
+
dynamic,cheap_q,2000,1337,ok,3720.43,123.64,4.4459,9250,4.4464,False,out-openwebtext_dynamic_warmup2000_seed1337_p1,
|
| 13 |
+
dynamic,cheap_q,2000,123,ok,3725.86,123.64,4.4066,10000,4.4066,True,out-openwebtext_dynamic_warmup2000_seed123_p1,
|
results/csv/priority2_runs_20260323_0623.csv
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,short_name,warmup,seed,status,runtime_sec,params_m,best_val,best_step,final_val,stable,out_dir,error
|
| 2 |
+
none,baseline_p2,2000,42,ok,3671.63,123.59,4.4705,9750,4.4839,False,out-openwebtext_none_warmup2000_seed42_p2,
|
| 3 |
+
none,baseline_p2,2000,1337,ok,3640.72,123.59,4.4591,9750,4.5028,False,out-openwebtext_none_warmup2000_seed1337_p2,
|
| 4 |
+
none,baseline_p2,2000,123,ok,3615.74,123.59,4.4768,9500,4.4972,False,out-openwebtext_none_warmup2000_seed123_p2,
|
| 5 |
+
dynamic_qa_conditioned_mlp128,mlp128_qa,2000,42,ok,3913.3,123.88,4.424,10000,4.424,True,out-openwebtext_dynamic_qa_conditioned_mlp128_warmup2000_seed42_p2,
|
| 6 |
+
dynamic_qa_conditioned_mlp128,mlp128_qa,2000,1337,ok,3928.35,123.88,4.3976,9750,4.4163,False,out-openwebtext_dynamic_qa_conditioned_mlp128_warmup2000_seed1337_p2,
|
| 7 |
+
dynamic_qa_conditioned_mlp128,mlp128_qa,2000,123,ok,3909.65,123.88,4.3879,9750,4.4036,False,out-openwebtext_dynamic_qa_conditioned_mlp128_warmup2000_seed123_p2,
|
| 8 |
+
static_gate_prehead,static_prehead,2000,42,ok,3750.43,123.6,4.4568,9750,4.471,False,out-openwebtext_static_gate_prehead_warmup2000_seed42_p2,
|
| 9 |
+
static_gate_prehead,static_prehead,2000,1337,failed,586.87,123.6,5.9137,1500,5.9137,False,out-openwebtext_static_gate_prehead_warmup2000_seed1337_p2,exit code 1
|
| 10 |
+
static_gate_prehead,static_prehead,2000,123,failed,114.3,123.6,6.9069,250,6.9069,False,out-openwebtext_static_gate_prehead_warmup2000_seed123_p2,exit code 1
|
| 11 |
+
dynamic_xconditioned_g1,full_x,2000,42,failed,119.3,130.67,6.9348,250,6.9348,False,out-openwebtext_dynamic_xconditioned_g1_warmup2000_seed42_p2,exit code 1
|
| 12 |
+
dynamic_xconditioned_g1,full_x,2000,1337,failed,119.55,130.67,6.937,250,6.937,False,out-openwebtext_dynamic_xconditioned_g1_warmup2000_seed1337_p2,exit code 1
|
| 13 |
+
dynamic_xconditioned_g1,full_x,2000,123,failed,119.47,130.67,6.9534,250,6.9534,False,out-openwebtext_dynamic_xconditioned_g1_warmup2000_seed123_p2,exit code 1
|
| 14 |
+
dynamic_x_g1_headshared_elementwise,x_headshared,2000,42,failed,115.91,123.64,6.9409,250,6.9409,False,out-openwebtext_dynamic_x_g1_headshared_elementwise_warmup2000_seed42_p2,exit code 1
|
| 15 |
+
dynamic_x_g1_headshared_elementwise,x_headshared,2000,1337,failed,116.86,123.64,6.9312,250,6.9312,False,out-openwebtext_dynamic_x_g1_headshared_elementwise_warmup2000_seed1337_p2,exit code 1
|
| 16 |
+
dynamic_x_g1_headshared_elementwise,x_headshared,2000,123,failed,116.7,123.64,6.9769,250,6.9769,False,out-openwebtext_dynamic_x_g1_headshared_elementwise_warmup2000_seed123_p2,exit code 1
|
results/csv/qa_screen_results.csv
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
variant,warmup,seed,status,runtime_sec,params_m,tokens_per_iter,best_val,best_step,final_val,final_step,out_dir,checkpoint_dir,log_path,error
|
| 2 |
+
dynamic_qa_conditioned,2000,42,ok,3871.08,123.69,"4,096",4.3807,10000,4.3807,10000,out-openwebtext_dynamic_qa_conditioned_wu2000_seed42,out-openwebtext_dynamic_qa_conditioned_wu2000_seed42,out-openwebtext_dynamic_qa_conditioned_wu2000_seed42\train.log,
|
| 3 |
+
dynamic_qa_conditioned_mlp128,2000,42,ok,3893.58,123.88,"4,096",4.4282,10000,4.4282,10000,out-openwebtext_dynamic_qa_conditioned_mlp128_wu2000_seed42,out-openwebtext_dynamic_qa_conditioned_mlp128_wu2000_seed42,out-openwebtext_dynamic_qa_conditioned_mlp128_wu2000_seed42\train.log,
|
| 4 |
+
dynamic_qconditioned_mlp192,2000,42,ok,3967.29,123.88,"4,096",4.4337,10000,4.4337,10000,out-openwebtext_dynamic_qconditioned_mlp192_wu2000_seed42,out-openwebtext_dynamic_qconditioned_mlp192_wu2000_seed42,out-openwebtext_dynamic_qconditioned_mlp192_wu2000_seed42\train.log,
|
| 5 |
+
dynamic_q_headshared_elementwise,2000,42,ok,3767.82,123.64,"4,096",4.4073,10000,4.4073,10000,out-openwebtext_dynamic_q_headshared_elementwise_wu2000_seed42,out-openwebtext_dynamic_q_headshared_elementwise_wu2000_seed42,out-openwebtext_dynamic_q_headshared_elementwise_wu2000_seed42\train.log,
|
| 6 |
+
dynamic_qa_headshared_elementwise,2000,42,ok,3782.7,123.69,"4,096",4.3844,10000,4.3844,10000,out-openwebtext_dynamic_qa_headshared_elementwise_wu2000_seed42,out-openwebtext_dynamic_qa_headshared_elementwise_wu2000_seed42,out-openwebtext_dynamic_qa_headshared_elementwise_wu2000_seed42\train.log,
|
| 7 |
+
dynamic_qa_conditioned,500,42,ok,3787.43,123.69,"4,096",4.6003,10000,4.6003,10000,out-openwebtext_dynamic_qa_conditioned_wu500_seed42,out-openwebtext_dynamic_qa_conditioned_wu500_seed42,out-openwebtext_dynamic_qa_conditioned_wu500_seed42\train.log,
|
| 8 |
+
dynamic_qa_conditioned_mlp128,500,42,ok,3919.42,123.88,"4,096",4.6761,10000,4.6761,10000,out-openwebtext_dynamic_qa_conditioned_mlp128_wu500_seed42,out-openwebtext_dynamic_qa_conditioned_mlp128_wu500_seed42,out-openwebtext_dynamic_qa_conditioned_mlp128_wu500_seed42\train.log,
|
| 9 |
+
dynamic_qconditioned_mlp192,500,42,ok,3922.5,123.88,"4,096",4.6548,9750,4.6548,10000,out-openwebtext_dynamic_qconditioned_mlp192_wu500_seed42,out-openwebtext_dynamic_qconditioned_mlp192_wu500_seed42,out-openwebtext_dynamic_qconditioned_mlp192_wu500_seed42\train.log,
|
| 10 |
+
dynamic_q_headshared_elementwise,500,42,ok,3768.24,123.64,"4,096",4.6539,10000,4.6539,10000,out-openwebtext_dynamic_q_headshared_elementwise_wu500_seed42,out-openwebtext_dynamic_q_headshared_elementwise_wu500_seed42,out-openwebtext_dynamic_q_headshared_elementwise_wu500_seed42\train.log,
|
| 11 |
+
dynamic_qa_headshared_elementwise,500,42,ok,3865.7,123.69,"4,096",4.617,10000,4.617,10000,out-openwebtext_dynamic_qa_headshared_elementwise_wu500_seed42,out-openwebtext_dynamic_qa_headshared_elementwise_wu500_seed42,out-openwebtext_dynamic_qa_headshared_elementwise_wu500_seed42\train.log,
|
results/csv/qv_124m_3seed_results_20260310_0041.csv
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
Variant,Seed,Final_Val_Loss
|
results/csv/qv_124m_5000_results_20260310_0900.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Final_Val_Loss,vs_Baseline
|
| 2 |
+
none,1.4982,+0.0000
|
| 3 |
+
dynamic,1.4879,-0.0103
|
results/csv/qv_3seed_results_20260310_0036.csv
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Seed,Final_Val_Loss
|
| 2 |
+
none,1337,1.6881
|
| 3 |
+
none,42,1.7212
|
| 4 |
+
none,123,1.6995
|
| 5 |
+
dynamic,1337,1.5365
|
| 6 |
+
dynamic,42,1.5008
|
| 7 |
+
dynamic,123,1.5318
|
results/csv/qv_long_results_20260309_2224.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Final_Val_Loss,vs_Baseline
|
| 2 |
+
none,1.6785,+0.0000
|
| 3 |
+
dynamic,1.5407,-0.1378
|
results/csv/qv_openwebtext_results_20260311_0830.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Final_Val_Loss,vs_Baseline
|
| 2 |
+
none,6.0341,+0.0000
|
| 3 |
+
qvnorm,5.8979,-0.1362
|
results/csv/qv_results_20260309_2137.csv
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Variant,Final_Val_Loss,vs_Baseline
|
| 2 |
+
none,1.5276,+0.0000
|
| 3 |
+
vnorm,1.5264,-0.0012
|
| 4 |
+
qvnorm,1.5347,+0.0071
|
| 5 |
+
dynamic,1.5118,-0.0158
|