Nicholas0228's picture
Complete original CR10 run with matched 1k, 4k and 7k checkpoints
76a191a verified
Raw History Blame Contribute Delete
35.4 kB
{
"checkpoint_count": 47,
"checkpoints": [
{
"clip_norm": 5.0,
"code_commit": "35ea8e70259242b072545bf0da61e8fc49e7101c",
"dataset_manifest_sha256": "154ff0781168c3b555f54415a1588f1315d530e316e4dbf8ff833572d16a23cc",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0001,
"optimizer_step": 1000,
"repo_path": "c1-1/cr10_staged_cosine/packed_temporal_grid/iter_000001000",
"sample_exposure": 8000,
"wandb_run_id": "g9m8yivu"
},
{
"clip_norm": 5.0,
"code_commit": "35ea8e70259242b072545bf0da61e8fc49e7101c",
"dataset_manifest_sha256": "154ff0781168c3b555f54415a1588f1315d530e316e4dbf8ff833572d16a23cc",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0001,
"optimizer_step": 2000,
"repo_path": "c1-1/cr10_staged_cosine/packed_temporal_grid/iter_000002000",
"sample_exposure": 16000,
"wandb_run_id": "g9m8yivu"
},
{
"clip_norm": 5.0,
"code_commit": "35ea8e70259242b072545bf0da61e8fc49e7101c",
"dataset_manifest_sha256": "154ff0781168c3b555f54415a1588f1315d530e316e4dbf8ff833572d16a23cc",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0001,
"optimizer_step": 3000,
"repo_path": "c1-1/cr10_staged_cosine/packed_temporal_grid/iter_000003000",
"sample_exposure": 24000,
"wandb_run_id": "g9m8yivu"
},
{
"clip_norm": 5.0,
"code_commit": "35ea8e70259242b072545bf0da61e8fc49e7101c",
"dataset_manifest_sha256": "154ff0781168c3b555f54415a1588f1315d530e316e4dbf8ff833572d16a23cc",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0001,
"optimizer_step": 4000,
"repo_path": "c1-1/cr10_staged_cosine/packed_temporal_grid/iter_000004000",
"sample_exposure": 32000,
"wandb_run_id": "g9m8yivu"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "428adedea71ae54cb30ebb6ac8c20a1083694bf6157a28d13b39dc555d870d79",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0002,
"optimizer_step": 3625,
"repo_path": "c1-1/packed_temporal_grid/iter_000003625",
"sample_exposure": 29000,
"wandb_run_id": "85efg1fg"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "428adedea71ae54cb30ebb6ac8c20a1083694bf6157a28d13b39dc555d870d79",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0002,
"optimizer_step": 4000,
"repo_path": "c1-1/packed_temporal_grid/iter_000004000",
"sample_exposure": 32000,
"wandb_run_id": "85efg1fg"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "428adedea71ae54cb30ebb6ac8c20a1083694bf6157a28d13b39dc555d870d79",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c1-1",
"optimizer_lr": 0.0002,
"optimizer_step": 5000,
"repo_path": "c1-1/packed_temporal_grid/iter_000005000",
"sample_exposure": 40000,
"wandb_run_id": "85efg1fg"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "428adedea71ae54cb30ebb6ac8c20a1083694bf6157a28d13b39dc555d870d79",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c1-1",
"optimizer_lr": 0.0002,
"optimizer_step": 4000,
"repo_path": "c1-1/source_aligned/iter_000004000",
"sample_exposure": 32000,
"wandb_run_id": "7pieax8v"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "428adedea71ae54cb30ebb6ac8c20a1083694bf6157a28d13b39dc555d870d79",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c1-1",
"optimizer_lr": 0.0002,
"optimizer_step": 4500,
"repo_path": "c1-1/source_aligned/iter_000004500",
"sample_exposure": 36000,
"wandb_run_id": "7pieax8v"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "428adedea71ae54cb30ebb6ac8c20a1083694bf6157a28d13b39dc555d870d79",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c1-1",
"optimizer_lr": 0.0002,
"optimizer_step": 5000,
"repo_path": "c1-1/source_aligned/iter_000005000",
"sample_exposure": 40000,
"wandb_run_id": "7pieax8v"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 1000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000001000",
"sample_exposure": 8000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 2000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000002000",
"sample_exposure": 16000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 3000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000003000",
"sample_exposure": 24000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 4000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000004000",
"sample_exposure": 32000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 5000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000005000",
"sample_exposure": 40000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 6000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000006000",
"sample_exposure": 48000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 7000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000007000",
"sample_exposure": 56000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 8000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000008000",
"sample_exposure": 64000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 9000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000009000",
"sample_exposure": 72000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"checkpoint_model_sha256": "1f60a5aebfe0b7305e052b81ca5a67cefc28fdab51b05bef5ceef65450dcdeec",
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"moviegen_vbench_long": {
"26s": {
"arm": "c3_periodic_cr4_packed_block_aligned",
"comparison_sha256": "5b03fa6782287753ac9e01edca6f157d6d89ec62b279a23567d942b3ef89fe0d",
"metrics": {
"aesthetic_quality": 0.5965064599777437,
"background_consistency": 0.9666898061357392,
"dynamic_degree": 0.46171874999999984,
"imaging_quality": 0.6536598342804102,
"motion_smoothness": 0.9879662829564213,
"raw_quality_macro": 0.7740303664483891,
"raw_temporal_macro": 0.848503976108045,
"raw_visual_macro": 0.625083147129077,
"subject_consistency": 0.9776410653400195,
"vbench_quality_subset_score": 0.7958231254936546
},
"protocol": "moviegen_videobench_c3_417frame_128x5_v1",
"sample_count": 640
},
"60s": {
"arm": "c3_periodic_cr4_packed_block_aligned",
"comparison_sha256": "6f15a44cb4db8c4d81695a9c5eb0b46978a8c6bf2ac3ac3257a686b3f5686da9",
"metrics": {
"aesthetic_quality": 0.5655917563188507,
"background_consistency": 0.9643949605061275,
"dynamic_degree": 0.42609375000000027,
"imaging_quality": 0.5963953730178768,
"motion_smoothness": 0.9878605960639982,
"raw_quality_macro": 0.752271223695733,
"raw_temporal_macro": 0.8379100532094176,
"raw_visual_macro": 0.5809935646683637,
"subject_consistency": 0.9732909062675444,
"vbench_quality_subset_score": 0.7749946300255505
},
"protocol": "moviegen_videobench_c3_957frame_128x5_v1",
"sample_count": 640
}
},
"optimizer_lr": 0.0002,
"optimizer_step": 10000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/packed_temporal_grid/iter_000010000",
"sample_exposure": 80000,
"validation_metrics": {
"checkpoint_step": 10000,
"normalized_velocity_mse": 0.008829873414261,
"scaled_velocity_loss": 1.6964355086126186,
"velocity_cosine": 0.995569415756553,
"wandb_run_path": "a-little-hoof/Memory_Compression/val-c3_3_fixed_noise_validation_20260804-c3-packed-cosine"
},
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "4jresx1i"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 1000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000001000",
"sample_exposure": 8000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 2000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000002000",
"sample_exposure": 16000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 3000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000003000",
"sample_exposure": 24000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 4000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000004000",
"sample_exposure": 32000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 5000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000005000",
"sample_exposure": 40000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 6000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000006000",
"sample_exposure": 48000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 7000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000007000",
"sample_exposure": 56000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 8000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000008000",
"sample_exposure": 64000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 9000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000009000",
"sample_exposure": 72000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"lr_schedule": {
"cosine_min_factor": 0.25,
"cycle_length_steps": 10000,
"f_max": 1.0,
"f_start": 1e-06,
"final_lr": 5e-05,
"peak_lr": 0.0002,
"scheduler_target": "rcm.utils.lr_scheduler.LambdaWarmUpCosineScheduler",
"type": "cosine",
"warmup_steps": 100
},
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 10000,
"repo_path": "c3-3/cosine_2e-4_to5e-5_clip5/source_aligned/iter_000010000",
"sample_exposure": 80000,
"variant": "cosine_2e-4_to5e-5_clip5",
"wandb_run_id": "jhnefd4q"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 2000,
"repo_path": "c3-3/packed_temporal_grid/iter_000002000",
"sample_exposure": 16000,
"wandb_run_id": "d43p7z74"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 4000,
"repo_path": "c3-3/packed_temporal_grid/iter_000004000",
"sample_exposure": 32000,
"wandb_run_id": "d43p7z74"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 5000,
"repo_path": "c3-3/packed_temporal_grid/iter_000005000",
"sample_exposure": 40000,
"wandb_run_id": "d43p7z74"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 6000,
"repo_path": "c3-3/packed_temporal_grid/iter_000006000",
"role": "failure_diagnostic",
"sample_exposure": 48000,
"wandb_run_id": "d43p7z74"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 10000,
"repo_path": "c3-3/packed_temporal_grid/iter_000010000",
"sample_exposure": 80000,
"wandb_run_id": "d43p7z74"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 4000,
"repo_path": "c3-3/source_aligned/iter_000004000",
"sample_exposure": 32000,
"wandb_run_id": "ynulkxtd"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 5000,
"repo_path": "c3-3/source_aligned/iter_000005000",
"sample_exposure": 40000,
"wandb_run_id": "ynulkxtd"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 6000,
"repo_path": "c3-3/source_aligned/iter_000006000",
"role": "stable_control",
"sample_exposure": 48000,
"wandb_run_id": "ynulkxtd"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 9000,
"repo_path": "c3-3/source_aligned/iter_000009000",
"sample_exposure": 72000,
"wandb_run_id": "ynulkxtd"
},
{
"clip_norm": 5.0,
"code_commit": "44f4c8b11f671e811f896539c5faa96db77a078d",
"dataset_manifest_sha256": "488feab621a47d5f5b77e1fb6547b24d64912cba66b7d6cababa53c1d4c8dd78",
"grad_clip": true,
"memory_rope_mode": "source_aligned",
"model_family": "c3-3",
"optimizer_lr": 0.0002,
"optimizer_step": 10000,
"repo_path": "c3-3/source_aligned/iter_000010000",
"sample_exposure": 80000,
"wandb_run_id": "ynulkxtd"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 1000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000001000",
"sample_exposure": 8000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 2000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000002000",
"sample_exposure": 16000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 3000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000003000",
"sample_exposure": 24000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 4000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000004000",
"sample_exposure": 32000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 5000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000005000",
"sample_exposure": 40000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 7000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000007000",
"sample_exposure": 56000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
},
{
"clip_norm": 5.0,
"compression_rate": 10,
"global_batch_size": 8,
"grad_clip": true,
"inference_code_commit": "757194f78bad10f50d660d14342a5f18b82470f0",
"inference_only": true,
"lr_schedule": "staged_cosine",
"memory_ratio": 0.1,
"memory_rope_mode": "packed_temporal_grid",
"model_family": "c3-3",
"optimizer_lr_peak": 0.0001,
"optimizer_step": 10000,
"repo_path": "c3-3/cr10_fullprefix_gradctxfix/iter_000010000",
"sample_exposure": 80000,
"training_objective": "cacheaware_future_self_forcing",
"training_run": "fullprefix_memoryonly_cr10_c3_3_gradctxfix_gbs8_lr1e4_stagedcosine_clip5_raw_10k",
"wandb_run_id": "sm6y0v1z"
}
]
}