Add 33 model file(s)
Browse files- t1_pure_satellite_xs_seed2/checkpoint-final.pth +3 -0
- t1_pure_satellite_xxs2_seed1/checkpoint-10Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed1/checkpoint-12Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed1/checkpoint-1Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed1/checkpoint-2Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed1/checkpoint-5Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed1/checkpoint-final.pth +3 -0
- t1_pure_satellite_xxs2_seed1/config.json +111 -0
- t1_pure_satellite_xxs2_seed2/checkpoint-10Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed2/checkpoint-12Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed2/checkpoint-1Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed2/checkpoint-2Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed2/checkpoint-5Blb.pth +3 -0
- t1_pure_satellite_xxs2_seed2/checkpoint-final.pth +3 -0
- t1_pure_satellite_xxs2_seed2/config.json +111 -0
- t1_pure_satellite_xxs3_seed1/checkpoint-10Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed1/checkpoint-12Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed1/checkpoint-1Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed1/checkpoint-2Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed1/checkpoint-5Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed1/checkpoint-final.pth +3 -0
- t1_pure_satellite_xxs3_seed1/config.json +111 -0
- t1_pure_satellite_xxs3_seed2/checkpoint-10Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed2/checkpoint-12Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed2/checkpoint-1Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed2/checkpoint-2Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed2/checkpoint-5Blb.pth +3 -0
- t1_pure_satellite_xxs3_seed2/checkpoint-final.pth +3 -0
- t1_pure_satellite_xxs3_seed2/config.json +111 -0
- t1_pure_text_xs_seed1/checkpoint-10Blb.pth +3 -0
- t1_pure_text_xs_seed1/checkpoint-12Blb.pth +3 -0
- t1_pure_text_xs_seed1/checkpoint-1Blb.pth +3 -0
- t1_pure_text_xs_seed1/config.json +110 -0
t1_pure_satellite_xs_seed2/checkpoint-final.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33e24e9e9e582776e8264ddc29d1e7e62211de028b053e1dac4afc4321c1895a
|
| 3 |
+
size 1083623728
|
t1_pure_satellite_xxs2_seed1/checkpoint-10Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aa8f419096995b6e9abe3a271abaf621ab8475f8e19d5ba8d839ae0df17b6596
|
| 3 |
+
size 571268720
|
t1_pure_satellite_xxs2_seed1/checkpoint-12Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:97387b291dfad5dbce84d3aeede24c02ec7e35189da04953f7e14cefe77f52d9
|
| 3 |
+
size 571268720
|
t1_pure_satellite_xxs2_seed1/checkpoint-1Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a47bff3187e8a6c442bb77f0155c0b804341dc09f76dbd679c006bb18406379
|
| 3 |
+
size 571268197
|
t1_pure_satellite_xxs2_seed1/checkpoint-2Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8011853380c64e73f3937405ae4aed628bc7d8218a6c99fb2df98a83e5114c7
|
| 3 |
+
size 571268197
|
t1_pure_satellite_xxs2_seed1/checkpoint-5Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:67881b20398f549f3d9857995f520b5f2fc4380e21fd92c3b7264d28064ff47c
|
| 3 |
+
size 571268197
|
t1_pure_satellite_xxs2_seed1/checkpoint-final.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a9aae518642b20d4860d3be9b6d5fc5f584d8e31c1b4d8f10496fc43b32a0aa
|
| 3 |
+
size 571268720
|
t1_pure_satellite_xxs2_seed1/config.json
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "t1_pure_satellite_xxs2_seed1",
|
| 3 |
+
"epoch": 0,
|
| 4 |
+
"n_params": 49007200,
|
| 5 |
+
"args": {
|
| 6 |
+
"run_name": "t1_pure_satellite_xxs2_seed1",
|
| 7 |
+
"batch_size": 32,
|
| 8 |
+
"epochs": 1,
|
| 9 |
+
"total_tokens": -1,
|
| 10 |
+
"lossbearing_tokens": 12,
|
| 11 |
+
"accum_iter": 32,
|
| 12 |
+
"effective_batch_size": null,
|
| 13 |
+
"lr_batch_scaling": "sqrt",
|
| 14 |
+
"save_ckpt_freq": 1,
|
| 15 |
+
"save_ckpt_freq_tokens": 5,
|
| 16 |
+
"debug_batch_stats": 0,
|
| 17 |
+
"model": "fm_xxs2_12d_swiglu_nobias",
|
| 18 |
+
"dtype": "bfloat16",
|
| 19 |
+
"loss_type": "token",
|
| 20 |
+
"context_length": 2048,
|
| 21 |
+
"pos_emb_ablation": "vanilla",
|
| 22 |
+
"max_blocks": 32,
|
| 23 |
+
"intra_pos_emb": "default",
|
| 24 |
+
"video_pos_3d": false,
|
| 25 |
+
"finetune": "",
|
| 26 |
+
"opt": "adamw",
|
| 27 |
+
"opt_eps": 1e-08,
|
| 28 |
+
"opt_betas": [
|
| 29 |
+
0.9,
|
| 30 |
+
0.95
|
| 31 |
+
],
|
| 32 |
+
"compute_grad_norm": true,
|
| 33 |
+
"clip_grad": 1.0,
|
| 34 |
+
"skip_grad": null,
|
| 35 |
+
"momentum": 0.9,
|
| 36 |
+
"weight_decay": 0.1,
|
| 37 |
+
"weight_decay_end": 0.1,
|
| 38 |
+
"blr": 0.001,
|
| 39 |
+
"min_blr": 5e-05,
|
| 40 |
+
"scheduler": "wsd",
|
| 41 |
+
"decay_fraction": 0.1,
|
| 42 |
+
"warmup_steps": 250,
|
| 43 |
+
"cooldown_steps": -1,
|
| 44 |
+
"data_config": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml",
|
| 45 |
+
"data_config_val": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml",
|
| 46 |
+
"cross_eval_val": "",
|
| 47 |
+
"s3_endpoint": "",
|
| 48 |
+
"s3_data_endpoint": "",
|
| 49 |
+
"s3_multipart_chunksize_mb": 512,
|
| 50 |
+
"s3_multipart_threshold_mb": 512,
|
| 51 |
+
"s3_max_io_queue": 100,
|
| 52 |
+
"text_tokenizer_path": "fourm/utils/tokenizer/trained/smollm_tokenizer.json",
|
| 53 |
+
"eval_freq": 1,
|
| 54 |
+
"eval_freq_tokens": 2,
|
| 55 |
+
"dist_eval": true,
|
| 56 |
+
"eval": false,
|
| 57 |
+
"val_sampling_mode": "without_replacement",
|
| 58 |
+
"output_dir": "output/seed/t1_pure_satellite_xxs2_seed1",
|
| 59 |
+
"device": "cuda",
|
| 60 |
+
"seed": 1,
|
| 61 |
+
"resume": "",
|
| 62 |
+
"auto_resume": true,
|
| 63 |
+
"start_epoch": 0,
|
| 64 |
+
"num_workers": 10,
|
| 65 |
+
"pin_mem": true,
|
| 66 |
+
"find_unused_params": true,
|
| 67 |
+
"rlimit": 4096,
|
| 68 |
+
"print_all": false,
|
| 69 |
+
"show_user_warnings": false,
|
| 70 |
+
"s3_save_dir": "",
|
| 71 |
+
"dist_url": "env://",
|
| 72 |
+
"log_wandb": true,
|
| 73 |
+
"wandb_project": "loom",
|
| 74 |
+
"wandb_entity": null,
|
| 75 |
+
"wandb_run_name": "t1_pure_satellite_xxs2_seed1",
|
| 76 |
+
"wandb_group": "t1-seedvar-xxs2",
|
| 77 |
+
"dim": null,
|
| 78 |
+
"depth": null,
|
| 79 |
+
"num_heads": null,
|
| 80 |
+
"mlp_ratio": null,
|
| 81 |
+
"log_grad_norms_every": 0,
|
| 82 |
+
"save_at_tokens": "1,2,5,10,12",
|
| 83 |
+
"unigram_entropies": null,
|
| 84 |
+
"cooldown_branch": false,
|
| 85 |
+
"cooldown_tokens": 0,
|
| 86 |
+
"config_path": "cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs2_seed1.yaml",
|
| 87 |
+
"rank": 0,
|
| 88 |
+
"world_size": 1,
|
| 89 |
+
"gpu": 0,
|
| 90 |
+
"distributed": true,
|
| 91 |
+
"dist_backend": "nccl",
|
| 92 |
+
"num_tasks": 1,
|
| 93 |
+
"sampling_mode": "without_replacement",
|
| 94 |
+
"max_micro_steps": 366211,
|
| 95 |
+
"lossbearing_target": 12000000000.0,
|
| 96 |
+
"lossbearing_mode": true,
|
| 97 |
+
"lr": 0.001,
|
| 98 |
+
"min_lr": 5e-05,
|
| 99 |
+
"global_micro_step": 366211,
|
| 100 |
+
"cum_lossbearing": 12000493568.0,
|
| 101 |
+
"cum_mod_tokens": {
|
| 102 |
+
"text": 92667904,
|
| 103 |
+
"tok_satellite@128": 11907825664
|
| 104 |
+
}
|
| 105 |
+
},
|
| 106 |
+
"source_configs": {
|
| 107 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs2_seed1.yaml": "# xxs2 \u03c3_seed variant of pure satellite (P1-ladder companion \u2014 the xxs2-scale training-seed\n# floor for the P1 sub-XS scaling anchors). Clone of run_xxs2_pure.yaml + seed 1;\n# seed0 = the existing output/p1/t1_pure_satellite_xxs2 anchor. Train-only self-cool \u2192 output/seed/.\n_base_: ../../../../models/dec/4m-xxs2-dec.yaml\nrun_name: t1_pure_satellite_xxs2_seed1\nseed: 1\ntotal_tokens: -1\nlossbearing_tokens: 12\ndata_config: cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml\ndata_config_val: cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml\nwarmup_steps: 250\neval_freq_tokens: 2\nsave_ckpt_freq_tokens: 5\nsave_at_tokens: 1,2,5,10,12\nwandb_group: t1-seedvar-xxs2\n",
|
| 108 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml": "# T1 pure unimodal training data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nepoch_size: 50000000\nsampler_type: temperature\ntemperature_tau: 1.0\nsources:\n- name: satellite_v2\n shards: data/shards/satellite_v2_reshuffled/train/shard_*.tar\n weight: 22408820\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben\n shards: data/shards/satellite_ben_shuffled/train/shard_{000000..000470}.tar\n weight: 470438\n modality_tokens:\n tok_satellite@128: 256\n format: npz\nsampling_mode: without_replacement\n",
|
| 109 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml": "# T1 pure unimodal validation data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nval_steps: 200\nsampler_type: uniform\nsources:\n- name: satellite_v2_val\n shards: data/shards/satellite_v2_reshuffled/val/shard_{000000..000049}.tar\n weight: 50000\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben_val\n shards: data/shards/satellite_ben_shuffled/val/shard_{000000..000009}.tar\n weight: 9600\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n"
|
| 110 |
+
}
|
| 111 |
+
}
|
t1_pure_satellite_xxs2_seed2/checkpoint-10Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60fb97cff4b555b4dac99fbd14a3ea6982bdbe272e28c4b94741e22b71a204e7
|
| 3 |
+
size 571268720
|
t1_pure_satellite_xxs2_seed2/checkpoint-12Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a255e99973b2f7aad8a65560f31612c13d9fd8c910cabdfc6521f35c043f6762
|
| 3 |
+
size 571268720
|
t1_pure_satellite_xxs2_seed2/checkpoint-1Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5aa5590a686f50e616c2b768e16218814497811ab367c72a782625e6842affcb
|
| 3 |
+
size 571268197
|
t1_pure_satellite_xxs2_seed2/checkpoint-2Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:67af3dbc48f11c0782fc5b2ef5910df2d44f512eb4ff4dfc6057226fe1063ad2
|
| 3 |
+
size 571268197
|
t1_pure_satellite_xxs2_seed2/checkpoint-5Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76cef0c753859c43c745ffaba44a4904a02019fd62fe2f97ffd3fdb88c4cf7b4
|
| 3 |
+
size 571268197
|
t1_pure_satellite_xxs2_seed2/checkpoint-final.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d74601f550fea003d9c47975dfcb8f56b673a2bcf7ab5b67527870cc71cd1e8c
|
| 3 |
+
size 571268720
|
t1_pure_satellite_xxs2_seed2/config.json
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "t1_pure_satellite_xxs2_seed2",
|
| 3 |
+
"epoch": 0,
|
| 4 |
+
"n_params": 49007200,
|
| 5 |
+
"args": {
|
| 6 |
+
"run_name": "t1_pure_satellite_xxs2_seed2",
|
| 7 |
+
"batch_size": 32,
|
| 8 |
+
"epochs": 1,
|
| 9 |
+
"total_tokens": -1,
|
| 10 |
+
"lossbearing_tokens": 12,
|
| 11 |
+
"accum_iter": 32,
|
| 12 |
+
"effective_batch_size": null,
|
| 13 |
+
"lr_batch_scaling": "sqrt",
|
| 14 |
+
"save_ckpt_freq": 1,
|
| 15 |
+
"save_ckpt_freq_tokens": 5,
|
| 16 |
+
"debug_batch_stats": 0,
|
| 17 |
+
"model": "fm_xxs2_12d_swiglu_nobias",
|
| 18 |
+
"dtype": "bfloat16",
|
| 19 |
+
"loss_type": "token",
|
| 20 |
+
"context_length": 2048,
|
| 21 |
+
"pos_emb_ablation": "vanilla",
|
| 22 |
+
"max_blocks": 32,
|
| 23 |
+
"intra_pos_emb": "default",
|
| 24 |
+
"video_pos_3d": false,
|
| 25 |
+
"finetune": "",
|
| 26 |
+
"opt": "adamw",
|
| 27 |
+
"opt_eps": 1e-08,
|
| 28 |
+
"opt_betas": [
|
| 29 |
+
0.9,
|
| 30 |
+
0.95
|
| 31 |
+
],
|
| 32 |
+
"compute_grad_norm": true,
|
| 33 |
+
"clip_grad": 1.0,
|
| 34 |
+
"skip_grad": null,
|
| 35 |
+
"momentum": 0.9,
|
| 36 |
+
"weight_decay": 0.1,
|
| 37 |
+
"weight_decay_end": 0.1,
|
| 38 |
+
"blr": 0.001,
|
| 39 |
+
"min_blr": 5e-05,
|
| 40 |
+
"scheduler": "wsd",
|
| 41 |
+
"decay_fraction": 0.1,
|
| 42 |
+
"warmup_steps": 250,
|
| 43 |
+
"cooldown_steps": -1,
|
| 44 |
+
"data_config": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml",
|
| 45 |
+
"data_config_val": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml",
|
| 46 |
+
"cross_eval_val": "",
|
| 47 |
+
"s3_endpoint": "",
|
| 48 |
+
"s3_data_endpoint": "",
|
| 49 |
+
"s3_multipart_chunksize_mb": 512,
|
| 50 |
+
"s3_multipart_threshold_mb": 512,
|
| 51 |
+
"s3_max_io_queue": 100,
|
| 52 |
+
"text_tokenizer_path": "fourm/utils/tokenizer/trained/smollm_tokenizer.json",
|
| 53 |
+
"eval_freq": 1,
|
| 54 |
+
"eval_freq_tokens": 2,
|
| 55 |
+
"dist_eval": true,
|
| 56 |
+
"eval": false,
|
| 57 |
+
"val_sampling_mode": "without_replacement",
|
| 58 |
+
"output_dir": "output/seed/t1_pure_satellite_xxs2_seed2",
|
| 59 |
+
"device": "cuda",
|
| 60 |
+
"seed": 2,
|
| 61 |
+
"resume": "",
|
| 62 |
+
"auto_resume": true,
|
| 63 |
+
"start_epoch": 0,
|
| 64 |
+
"num_workers": 10,
|
| 65 |
+
"pin_mem": true,
|
| 66 |
+
"find_unused_params": true,
|
| 67 |
+
"rlimit": 4096,
|
| 68 |
+
"print_all": false,
|
| 69 |
+
"show_user_warnings": false,
|
| 70 |
+
"s3_save_dir": "",
|
| 71 |
+
"dist_url": "env://",
|
| 72 |
+
"log_wandb": true,
|
| 73 |
+
"wandb_project": "loom",
|
| 74 |
+
"wandb_entity": null,
|
| 75 |
+
"wandb_run_name": "t1_pure_satellite_xxs2_seed2",
|
| 76 |
+
"wandb_group": "t1-seedvar-xxs2",
|
| 77 |
+
"dim": null,
|
| 78 |
+
"depth": null,
|
| 79 |
+
"num_heads": null,
|
| 80 |
+
"mlp_ratio": null,
|
| 81 |
+
"log_grad_norms_every": 0,
|
| 82 |
+
"save_at_tokens": "1,2,5,10,12",
|
| 83 |
+
"unigram_entropies": null,
|
| 84 |
+
"cooldown_branch": false,
|
| 85 |
+
"cooldown_tokens": 0,
|
| 86 |
+
"config_path": "cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs2_seed2.yaml",
|
| 87 |
+
"rank": 0,
|
| 88 |
+
"world_size": 1,
|
| 89 |
+
"gpu": 0,
|
| 90 |
+
"distributed": true,
|
| 91 |
+
"dist_backend": "nccl",
|
| 92 |
+
"num_tasks": 1,
|
| 93 |
+
"sampling_mode": "without_replacement",
|
| 94 |
+
"max_micro_steps": 366211,
|
| 95 |
+
"lossbearing_target": 12000000000.0,
|
| 96 |
+
"lossbearing_mode": true,
|
| 97 |
+
"lr": 0.001,
|
| 98 |
+
"min_lr": 5e-05,
|
| 99 |
+
"global_micro_step": 366211,
|
| 100 |
+
"cum_lossbearing": 12000493568.0,
|
| 101 |
+
"cum_mod_tokens": {
|
| 102 |
+
"text": 92667904,
|
| 103 |
+
"tok_satellite@128": 11907825664
|
| 104 |
+
}
|
| 105 |
+
},
|
| 106 |
+
"source_configs": {
|
| 107 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs2_seed2.yaml": "# xxs2 \u03c3_seed variant of pure satellite (P1-ladder companion \u2014 the xxs2-scale training-seed\n# floor for the P1 sub-XS scaling anchors). Clone of run_xxs2_pure.yaml + seed 2;\n# seed0 = the existing output/p1/t1_pure_satellite_xxs2 anchor. Train-only self-cool \u2192 output/seed/.\n_base_: ../../../../models/dec/4m-xxs2-dec.yaml\nrun_name: t1_pure_satellite_xxs2_seed2\nseed: 2\ntotal_tokens: -1\nlossbearing_tokens: 12\ndata_config: cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml\ndata_config_val: cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml\nwarmup_steps: 250\neval_freq_tokens: 2\nsave_ckpt_freq_tokens: 5\nsave_at_tokens: 1,2,5,10,12\nwandb_group: t1-seedvar-xxs2\n",
|
| 108 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml": "# T1 pure unimodal training data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nepoch_size: 50000000\nsampler_type: temperature\ntemperature_tau: 1.0\nsources:\n- name: satellite_v2\n shards: data/shards/satellite_v2_reshuffled/train/shard_*.tar\n weight: 22408820\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben\n shards: data/shards/satellite_ben_shuffled/train/shard_{000000..000470}.tar\n weight: 470438\n modality_tokens:\n tok_satellite@128: 256\n format: npz\nsampling_mode: without_replacement\n",
|
| 109 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml": "# T1 pure unimodal validation data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nval_steps: 200\nsampler_type: uniform\nsources:\n- name: satellite_v2_val\n shards: data/shards/satellite_v2_reshuffled/val/shard_{000000..000049}.tar\n weight: 50000\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben_val\n shards: data/shards/satellite_ben_shuffled/val/shard_{000000..000009}.tar\n weight: 9600\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n"
|
| 110 |
+
}
|
| 111 |
+
}
|
t1_pure_satellite_xxs3_seed1/checkpoint-10Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:033848de3c7365621df1547c26df15cb707c60aa3348be91da22e1501d714767
|
| 3 |
+
size 727953264
|
t1_pure_satellite_xxs3_seed1/checkpoint-12Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8dd4e00e655bcc350795a3c20e1965f5862f18492e9847466880f7da2657df28
|
| 3 |
+
size 727953264
|
t1_pure_satellite_xxs3_seed1/checkpoint-1Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7f2311af71a27f96500ecc4162e5ebdac1e1170fccc0538ca82cd6594c1c772
|
| 3 |
+
size 727952741
|
t1_pure_satellite_xxs3_seed1/checkpoint-2Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e935e835722bbb620d4dd4076eee1e0c71abe64de19d1ecb400be126dccad52
|
| 3 |
+
size 727952741
|
t1_pure_satellite_xxs3_seed1/checkpoint-5Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a99c41d25e266d29a1d5034f85a183f86477d134446097655919439e92373415
|
| 3 |
+
size 727952741
|
t1_pure_satellite_xxs3_seed1/checkpoint-final.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fdc569ea4d86368732b9c07adf9b1dcb0e06c33f8cf76b213965da85fb2e7210
|
| 3 |
+
size 727953264
|
t1_pure_satellite_xxs3_seed1/config.json
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "t1_pure_satellite_xxs3_seed1",
|
| 3 |
+
"epoch": 0,
|
| 4 |
+
"n_params": 62338464,
|
| 5 |
+
"args": {
|
| 6 |
+
"run_name": "t1_pure_satellite_xxs3_seed1",
|
| 7 |
+
"batch_size": 32,
|
| 8 |
+
"epochs": 1,
|
| 9 |
+
"total_tokens": -1,
|
| 10 |
+
"lossbearing_tokens": 12,
|
| 11 |
+
"accum_iter": 32,
|
| 12 |
+
"effective_batch_size": null,
|
| 13 |
+
"lr_batch_scaling": "sqrt",
|
| 14 |
+
"save_ckpt_freq": 1,
|
| 15 |
+
"save_ckpt_freq_tokens": 5,
|
| 16 |
+
"debug_batch_stats": 0,
|
| 17 |
+
"model": "fm_xxs3_12d_swiglu_nobias",
|
| 18 |
+
"dtype": "bfloat16",
|
| 19 |
+
"loss_type": "token",
|
| 20 |
+
"context_length": 2048,
|
| 21 |
+
"pos_emb_ablation": "vanilla",
|
| 22 |
+
"max_blocks": 32,
|
| 23 |
+
"intra_pos_emb": "default",
|
| 24 |
+
"video_pos_3d": false,
|
| 25 |
+
"finetune": "",
|
| 26 |
+
"opt": "adamw",
|
| 27 |
+
"opt_eps": 1e-08,
|
| 28 |
+
"opt_betas": [
|
| 29 |
+
0.9,
|
| 30 |
+
0.95
|
| 31 |
+
],
|
| 32 |
+
"compute_grad_norm": true,
|
| 33 |
+
"clip_grad": 1.0,
|
| 34 |
+
"skip_grad": null,
|
| 35 |
+
"momentum": 0.9,
|
| 36 |
+
"weight_decay": 0.1,
|
| 37 |
+
"weight_decay_end": 0.1,
|
| 38 |
+
"blr": 0.001,
|
| 39 |
+
"min_blr": 5e-05,
|
| 40 |
+
"scheduler": "wsd",
|
| 41 |
+
"decay_fraction": 0.1,
|
| 42 |
+
"warmup_steps": 250,
|
| 43 |
+
"cooldown_steps": -1,
|
| 44 |
+
"data_config": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml",
|
| 45 |
+
"data_config_val": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml",
|
| 46 |
+
"cross_eval_val": "",
|
| 47 |
+
"s3_endpoint": "",
|
| 48 |
+
"s3_data_endpoint": "",
|
| 49 |
+
"s3_multipart_chunksize_mb": 512,
|
| 50 |
+
"s3_multipart_threshold_mb": 512,
|
| 51 |
+
"s3_max_io_queue": 100,
|
| 52 |
+
"text_tokenizer_path": "fourm/utils/tokenizer/trained/smollm_tokenizer.json",
|
| 53 |
+
"eval_freq": 1,
|
| 54 |
+
"eval_freq_tokens": 2,
|
| 55 |
+
"dist_eval": true,
|
| 56 |
+
"eval": false,
|
| 57 |
+
"val_sampling_mode": "without_replacement",
|
| 58 |
+
"output_dir": "output/seed/t1_pure_satellite_xxs3_seed1",
|
| 59 |
+
"device": "cuda",
|
| 60 |
+
"seed": 1,
|
| 61 |
+
"resume": "",
|
| 62 |
+
"auto_resume": true,
|
| 63 |
+
"start_epoch": 0,
|
| 64 |
+
"num_workers": 10,
|
| 65 |
+
"pin_mem": true,
|
| 66 |
+
"find_unused_params": true,
|
| 67 |
+
"rlimit": 4096,
|
| 68 |
+
"print_all": false,
|
| 69 |
+
"show_user_warnings": false,
|
| 70 |
+
"s3_save_dir": "",
|
| 71 |
+
"dist_url": "env://",
|
| 72 |
+
"log_wandb": true,
|
| 73 |
+
"wandb_project": "loom",
|
| 74 |
+
"wandb_entity": null,
|
| 75 |
+
"wandb_run_name": "t1_pure_satellite_xxs3_seed1",
|
| 76 |
+
"wandb_group": "t1-seedvar-xxs3",
|
| 77 |
+
"dim": null,
|
| 78 |
+
"depth": null,
|
| 79 |
+
"num_heads": null,
|
| 80 |
+
"mlp_ratio": null,
|
| 81 |
+
"log_grad_norms_every": 0,
|
| 82 |
+
"save_at_tokens": "1,2,5,10,12",
|
| 83 |
+
"unigram_entropies": null,
|
| 84 |
+
"cooldown_branch": false,
|
| 85 |
+
"cooldown_tokens": 0,
|
| 86 |
+
"config_path": "cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs3_seed1.yaml",
|
| 87 |
+
"rank": 0,
|
| 88 |
+
"world_size": 1,
|
| 89 |
+
"gpu": 0,
|
| 90 |
+
"distributed": true,
|
| 91 |
+
"dist_backend": "nccl",
|
| 92 |
+
"num_tasks": 1,
|
| 93 |
+
"sampling_mode": "without_replacement",
|
| 94 |
+
"max_micro_steps": 366211,
|
| 95 |
+
"lossbearing_target": 12000000000.0,
|
| 96 |
+
"lossbearing_mode": true,
|
| 97 |
+
"lr": 0.001,
|
| 98 |
+
"min_lr": 5e-05,
|
| 99 |
+
"global_micro_step": 366211,
|
| 100 |
+
"cum_lossbearing": 12000493568.0,
|
| 101 |
+
"cum_mod_tokens": {
|
| 102 |
+
"text": 92667904,
|
| 103 |
+
"tok_satellite@128": 11907825664
|
| 104 |
+
}
|
| 105 |
+
},
|
| 106 |
+
"source_configs": {
|
| 107 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs3_seed1.yaml": "# xxs3 \u03c3_seed variant of pure satellite (P1-ladder companion \u2014 the xxs3-scale training-seed\n# floor for the P1 sub-XS scaling anchors). Clone of run_xxs3_pure.yaml + seed 1;\n# seed0 = the existing output/p1/t1_pure_satellite_xxs3 anchor. Train-only self-cool \u2192 output/seed/.\n_base_: ../../../../models/dec/4m-xxs3-dec.yaml\nrun_name: t1_pure_satellite_xxs3_seed1\nseed: 1\ntotal_tokens: -1\nlossbearing_tokens: 12\ndata_config: cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml\ndata_config_val: cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml\nwarmup_steps: 250\neval_freq_tokens: 2\nsave_ckpt_freq_tokens: 5\nsave_at_tokens: 1,2,5,10,12\nwandb_group: t1-seedvar-xxs3\n",
|
| 108 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml": "# T1 pure unimodal training data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nepoch_size: 50000000\nsampler_type: temperature\ntemperature_tau: 1.0\nsources:\n- name: satellite_v2\n shards: data/shards/satellite_v2_reshuffled/train/shard_*.tar\n weight: 22408820\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben\n shards: data/shards/satellite_ben_shuffled/train/shard_{000000..000470}.tar\n weight: 470438\n modality_tokens:\n tok_satellite@128: 256\n format: npz\nsampling_mode: without_replacement\n",
|
| 109 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml": "# T1 pure unimodal validation data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nval_steps: 200\nsampler_type: uniform\nsources:\n- name: satellite_v2_val\n shards: data/shards/satellite_v2_reshuffled/val/shard_{000000..000049}.tar\n weight: 50000\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben_val\n shards: data/shards/satellite_ben_shuffled/val/shard_{000000..000009}.tar\n weight: 9600\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n"
|
| 110 |
+
}
|
| 111 |
+
}
|
t1_pure_satellite_xxs3_seed2/checkpoint-10Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f06e3bab3dc73131a9b0c3daf741268036610a5eada7c2937786c4f7407b881
|
| 3 |
+
size 727953264
|
t1_pure_satellite_xxs3_seed2/checkpoint-12Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f55c0d1db0f4b729733861e57c119163f92637fbec6ec0c9d884936a6c7980f
|
| 3 |
+
size 727953264
|
t1_pure_satellite_xxs3_seed2/checkpoint-1Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:adf27f8bf983e047e2007968a51a68178468823d278ec26ed5a70157083f69b7
|
| 3 |
+
size 727952741
|
t1_pure_satellite_xxs3_seed2/checkpoint-2Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1b91b4f71ae2321890a4616bd9209c5c653a73abe536e8b30c90aadcd0134639
|
| 3 |
+
size 727952741
|
t1_pure_satellite_xxs3_seed2/checkpoint-5Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbefce48c97f5a7392366abd3f4c8d4503335c2e0c574cfb1daa267f7a44f7f3
|
| 3 |
+
size 727952741
|
t1_pure_satellite_xxs3_seed2/checkpoint-final.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0304f9f74688fac1bd3337350ae23e3902c5a4ea90919dd8238a57f1d1d9492
|
| 3 |
+
size 727953264
|
t1_pure_satellite_xxs3_seed2/config.json
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "t1_pure_satellite_xxs3_seed2",
|
| 3 |
+
"epoch": 0,
|
| 4 |
+
"n_params": 62338464,
|
| 5 |
+
"args": {
|
| 6 |
+
"run_name": "t1_pure_satellite_xxs3_seed2",
|
| 7 |
+
"batch_size": 32,
|
| 8 |
+
"epochs": 1,
|
| 9 |
+
"total_tokens": -1,
|
| 10 |
+
"lossbearing_tokens": 12,
|
| 11 |
+
"accum_iter": 32,
|
| 12 |
+
"effective_batch_size": null,
|
| 13 |
+
"lr_batch_scaling": "sqrt",
|
| 14 |
+
"save_ckpt_freq": 1,
|
| 15 |
+
"save_ckpt_freq_tokens": 5,
|
| 16 |
+
"debug_batch_stats": 0,
|
| 17 |
+
"model": "fm_xxs3_12d_swiglu_nobias",
|
| 18 |
+
"dtype": "bfloat16",
|
| 19 |
+
"loss_type": "token",
|
| 20 |
+
"context_length": 2048,
|
| 21 |
+
"pos_emb_ablation": "vanilla",
|
| 22 |
+
"max_blocks": 32,
|
| 23 |
+
"intra_pos_emb": "default",
|
| 24 |
+
"video_pos_3d": false,
|
| 25 |
+
"finetune": "",
|
| 26 |
+
"opt": "adamw",
|
| 27 |
+
"opt_eps": 1e-08,
|
| 28 |
+
"opt_betas": [
|
| 29 |
+
0.9,
|
| 30 |
+
0.95
|
| 31 |
+
],
|
| 32 |
+
"compute_grad_norm": true,
|
| 33 |
+
"clip_grad": 1.0,
|
| 34 |
+
"skip_grad": null,
|
| 35 |
+
"momentum": 0.9,
|
| 36 |
+
"weight_decay": 0.1,
|
| 37 |
+
"weight_decay_end": 0.1,
|
| 38 |
+
"blr": 0.001,
|
| 39 |
+
"min_blr": 5e-05,
|
| 40 |
+
"scheduler": "wsd",
|
| 41 |
+
"decay_fraction": 0.1,
|
| 42 |
+
"warmup_steps": 250,
|
| 43 |
+
"cooldown_steps": -1,
|
| 44 |
+
"data_config": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml",
|
| 45 |
+
"data_config_val": "cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml",
|
| 46 |
+
"cross_eval_val": "",
|
| 47 |
+
"s3_endpoint": "",
|
| 48 |
+
"s3_data_endpoint": "",
|
| 49 |
+
"s3_multipart_chunksize_mb": 512,
|
| 50 |
+
"s3_multipart_threshold_mb": 512,
|
| 51 |
+
"s3_max_io_queue": 100,
|
| 52 |
+
"text_tokenizer_path": "fourm/utils/tokenizer/trained/smollm_tokenizer.json",
|
| 53 |
+
"eval_freq": 1,
|
| 54 |
+
"eval_freq_tokens": 2,
|
| 55 |
+
"dist_eval": true,
|
| 56 |
+
"eval": false,
|
| 57 |
+
"val_sampling_mode": "without_replacement",
|
| 58 |
+
"output_dir": "output/seed/t1_pure_satellite_xxs3_seed2",
|
| 59 |
+
"device": "cuda",
|
| 60 |
+
"seed": 2,
|
| 61 |
+
"resume": "",
|
| 62 |
+
"auto_resume": true,
|
| 63 |
+
"start_epoch": 0,
|
| 64 |
+
"num_workers": 10,
|
| 65 |
+
"pin_mem": true,
|
| 66 |
+
"find_unused_params": true,
|
| 67 |
+
"rlimit": 4096,
|
| 68 |
+
"print_all": false,
|
| 69 |
+
"show_user_warnings": false,
|
| 70 |
+
"s3_save_dir": "",
|
| 71 |
+
"dist_url": "env://",
|
| 72 |
+
"log_wandb": true,
|
| 73 |
+
"wandb_project": "loom",
|
| 74 |
+
"wandb_entity": null,
|
| 75 |
+
"wandb_run_name": "t1_pure_satellite_xxs3_seed2",
|
| 76 |
+
"wandb_group": "t1-seedvar-xxs3",
|
| 77 |
+
"dim": null,
|
| 78 |
+
"depth": null,
|
| 79 |
+
"num_heads": null,
|
| 80 |
+
"mlp_ratio": null,
|
| 81 |
+
"log_grad_norms_every": 0,
|
| 82 |
+
"save_at_tokens": "1,2,5,10,12",
|
| 83 |
+
"unigram_entropies": null,
|
| 84 |
+
"cooldown_branch": false,
|
| 85 |
+
"cooldown_tokens": 0,
|
| 86 |
+
"config_path": "cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs3_seed2.yaml",
|
| 87 |
+
"rank": 0,
|
| 88 |
+
"world_size": 1,
|
| 89 |
+
"gpu": 0,
|
| 90 |
+
"distributed": true,
|
| 91 |
+
"dist_backend": "nccl",
|
| 92 |
+
"num_tasks": 1,
|
| 93 |
+
"sampling_mode": "without_replacement",
|
| 94 |
+
"max_micro_steps": 366211,
|
| 95 |
+
"lossbearing_target": 12000000000.0,
|
| 96 |
+
"lossbearing_mode": true,
|
| 97 |
+
"lr": 0.001,
|
| 98 |
+
"min_lr": 5e-05,
|
| 99 |
+
"global_micro_step": 366211,
|
| 100 |
+
"cum_lossbearing": 12000493568.0,
|
| 101 |
+
"cum_mod_tokens": {
|
| 102 |
+
"text": 92667904,
|
| 103 |
+
"tok_satellite@128": 11907825664
|
| 104 |
+
}
|
| 105 |
+
},
|
| 106 |
+
"source_configs": {
|
| 107 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/run_xxs3_seed2.yaml": "# xxs3 \u03c3_seed variant of pure satellite (P1-ladder companion \u2014 the xxs3-scale training-seed\n# floor for the P1 sub-XS scaling anchors). Clone of run_xxs3_pure.yaml + seed 2;\n# seed0 = the existing output/p1/t1_pure_satellite_xxs3 anchor. Train-only self-cool \u2192 output/seed/.\n_base_: ../../../../models/dec/4m-xxs3-dec.yaml\nrun_name: t1_pure_satellite_xxs3_seed2\nseed: 2\ntotal_tokens: -1\nlossbearing_tokens: 12\ndata_config: cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml\ndata_config_val: cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml\nwarmup_steps: 250\neval_freq_tokens: 2\nsave_ckpt_freq_tokens: 5\nsave_at_tokens: 1,2,5,10,12\nwandb_group: t1-seedvar-xxs3\n",
|
| 108 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_train_pure.yaml": "# T1 pure unimodal training data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nepoch_size: 50000000\nsampler_type: temperature\ntemperature_tau: 1.0\nsources:\n- name: satellite_v2\n shards: data/shards/satellite_v2_reshuffled/train/shard_*.tar\n weight: 22408820\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben\n shards: data/shards/satellite_ben_shuffled/train/shard_{000000..000470}.tar\n weight: 470438\n modality_tokens:\n tok_satellite@128: 256\n format: npz\nsampling_mode: without_replacement\n",
|
| 109 |
+
"cfgs/default/4m/experiments/t1/unimodal/satellite/data_val_pure.yaml": "# T1 pure unimodal validation data \u2014 satellite\n# Text stripped from all sources (pure modality-only condition)\n\nmodalities:\n- tok_satellite@128\nval_steps: 200\nsampler_type: uniform\nsources:\n- name: satellite_v2_val\n shards: data/shards/satellite_v2_reshuffled/val/shard_{000000..000049}.tar\n weight: 50000\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n- name: satellite_ben_val\n shards: data/shards/satellite_ben_shuffled/val/shard_{000000..000009}.tar\n weight: 9600\n modality_tokens:\n tok_satellite@128: 256\n format: npz\n"
|
| 110 |
+
}
|
| 111 |
+
}
|
t1_pure_text_xs_seed1/checkpoint-10Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:30244d937a25ad7a92f5848519258b8f8d44bf9f6a775e5ee4e03d596d8273c2
|
| 3 |
+
size 1070037090
|
t1_pure_text_xs_seed1/checkpoint-12Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c7f3b204314d94f9e202664724aac44854ce87263758b48169c03b06e15901a
|
| 3 |
+
size 1070037090
|
t1_pure_text_xs_seed1/checkpoint-1Blb.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0163d0536010a81f9fe0707ccbba3c88d2983ef9ddad1839e7cbc3c849c9bf24
|
| 3 |
+
size 1070036521
|
t1_pure_text_xs_seed1/config.json
ADDED
|
@@ -0,0 +1,110 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "t1_pure_text_xs_seed1",
|
| 3 |
+
"epoch": 0,
|
| 4 |
+
"n_params": 91262976,
|
| 5 |
+
"args": {
|
| 6 |
+
"run_name": "t1_pure_text_xs_seed1",
|
| 7 |
+
"batch_size": 32,
|
| 8 |
+
"epochs": 1,
|
| 9 |
+
"total_tokens": -1,
|
| 10 |
+
"lossbearing_tokens": 12,
|
| 11 |
+
"accum_iter": 32,
|
| 12 |
+
"effective_batch_size": null,
|
| 13 |
+
"lr_batch_scaling": "sqrt",
|
| 14 |
+
"save_ckpt_freq": 1,
|
| 15 |
+
"save_ckpt_freq_tokens": 5,
|
| 16 |
+
"debug_batch_stats": 0,
|
| 17 |
+
"model": "fm_xs_12d_swiglu_nobias",
|
| 18 |
+
"dtype": "bfloat16",
|
| 19 |
+
"loss_type": "token",
|
| 20 |
+
"context_length": 2048,
|
| 21 |
+
"pos_emb_ablation": "vanilla",
|
| 22 |
+
"max_blocks": 32,
|
| 23 |
+
"intra_pos_emb": "default",
|
| 24 |
+
"video_pos_3d": false,
|
| 25 |
+
"finetune": "",
|
| 26 |
+
"opt": "adamw",
|
| 27 |
+
"opt_eps": 1e-08,
|
| 28 |
+
"opt_betas": [
|
| 29 |
+
0.9,
|
| 30 |
+
0.95
|
| 31 |
+
],
|
| 32 |
+
"compute_grad_norm": true,
|
| 33 |
+
"clip_grad": 1.0,
|
| 34 |
+
"skip_grad": null,
|
| 35 |
+
"momentum": 0.9,
|
| 36 |
+
"weight_decay": 0.1,
|
| 37 |
+
"weight_decay_end": 0.1,
|
| 38 |
+
"blr": 0.001,
|
| 39 |
+
"min_blr": 5e-05,
|
| 40 |
+
"scheduler": "wsd",
|
| 41 |
+
"decay_fraction": 0.1,
|
| 42 |
+
"warmup_steps": 250,
|
| 43 |
+
"cooldown_steps": -1,
|
| 44 |
+
"data_config": "cfgs/default/4m/experiments/t1/unimodal/text/data_train.yaml",
|
| 45 |
+
"data_config_val": "cfgs/default/4m/experiments/t1/unimodal/text/data_val.yaml",
|
| 46 |
+
"cross_eval_val": "",
|
| 47 |
+
"s3_endpoint": "",
|
| 48 |
+
"s3_data_endpoint": "",
|
| 49 |
+
"s3_multipart_chunksize_mb": 512,
|
| 50 |
+
"s3_multipart_threshold_mb": 512,
|
| 51 |
+
"s3_max_io_queue": 100,
|
| 52 |
+
"text_tokenizer_path": "fourm/utils/tokenizer/trained/smollm_tokenizer.json",
|
| 53 |
+
"eval_freq": 1,
|
| 54 |
+
"eval_freq_tokens": 2,
|
| 55 |
+
"dist_eval": true,
|
| 56 |
+
"eval": false,
|
| 57 |
+
"val_sampling_mode": "without_replacement",
|
| 58 |
+
"output_dir": "output/seed/t1_pure_text_xs_seed1",
|
| 59 |
+
"device": "cuda",
|
| 60 |
+
"seed": 1,
|
| 61 |
+
"resume": "",
|
| 62 |
+
"auto_resume": true,
|
| 63 |
+
"start_epoch": 0,
|
| 64 |
+
"num_workers": 10,
|
| 65 |
+
"pin_mem": true,
|
| 66 |
+
"find_unused_params": true,
|
| 67 |
+
"rlimit": 4096,
|
| 68 |
+
"print_all": false,
|
| 69 |
+
"show_user_warnings": false,
|
| 70 |
+
"s3_save_dir": "",
|
| 71 |
+
"dist_url": "env://",
|
| 72 |
+
"log_wandb": true,
|
| 73 |
+
"wandb_project": "loom",
|
| 74 |
+
"wandb_entity": null,
|
| 75 |
+
"wandb_run_name": "t1_pure_text_xs_seed1",
|
| 76 |
+
"wandb_group": "t1-seedvar-xs",
|
| 77 |
+
"dim": null,
|
| 78 |
+
"depth": null,
|
| 79 |
+
"num_heads": null,
|
| 80 |
+
"mlp_ratio": null,
|
| 81 |
+
"log_grad_norms_every": 0,
|
| 82 |
+
"save_at_tokens": "1,2,5,10,12",
|
| 83 |
+
"unigram_entropies": null,
|
| 84 |
+
"cooldown_branch": false,
|
| 85 |
+
"cooldown_tokens": 0,
|
| 86 |
+
"config_path": "cfgs/default/4m/experiments/t1/unimodal/text/run_xs_seed1.yaml",
|
| 87 |
+
"rank": 0,
|
| 88 |
+
"world_size": 1,
|
| 89 |
+
"gpu": 0,
|
| 90 |
+
"distributed": true,
|
| 91 |
+
"dist_backend": "nccl",
|
| 92 |
+
"num_tasks": 1,
|
| 93 |
+
"sampling_mode": "without_replacement",
|
| 94 |
+
"max_micro_steps": 366211,
|
| 95 |
+
"lossbearing_target": 12000000000.0,
|
| 96 |
+
"lossbearing_mode": true,
|
| 97 |
+
"lr": 0.001,
|
| 98 |
+
"min_lr": 5e-05,
|
| 99 |
+
"global_micro_step": 366211,
|
| 100 |
+
"cum_lossbearing": 12001274853.0,
|
| 101 |
+
"cum_mod_tokens": {
|
| 102 |
+
"text": 12001274853
|
| 103 |
+
}
|
| 104 |
+
},
|
| 105 |
+
"source_configs": {
|
| 106 |
+
"cfgs/default/4m/experiments/t1/unimodal/text/run_xs_seed1.yaml": "# XS \u03c3_seed variant of pure text (Lever-B companion \u2014 the XS-scale training-seed\n# floor for the XS bimodal N-scan output/xs/). Clone of run_xs.yaml + seed 1;\n# seed0 = the existing output/lb/t1_pure_text_xs anchor. Train-only self-cool \u2192 output/seed/.\n_base_: ../../../../models/dec/4m-xs-dec.yaml\nrun_name: t1_pure_text_xs_seed1\nseed: 1\ntotal_tokens: -1\nlossbearing_tokens: 12\ndata_config: cfgs/default/4m/experiments/t1/unimodal/text/data_train.yaml\ndata_config_val: cfgs/default/4m/experiments/t1/unimodal/text/data_val.yaml\nwarmup_steps: 250\neval_freq_tokens: 2\nsave_ckpt_freq_tokens: 5\nsave_at_tokens: 1,2,5,10,12\nwandb_group: t1-seedvar-xs\n",
|
| 107 |
+
"cfgs/default/4m/experiments/t1/unimodal/text/data_train.yaml": "# T1 unimodal training data \u2014 text\n# Sources: dclm\n\nmodalities:\n- text\nepoch_size: 50000000\nsampler_type: temperature\ntemperature_tau: 1.0\nsources:\n- name: dclm\n shards: data/shards/dclm_reshuffled/train/shard_*.tar\n weight: 426009612\n modality_tokens:\n text: 1300\n format: npz\n",
|
| 108 |
+
"cfgs/default/4m/experiments/t1/unimodal/text/data_val.yaml": "# T1 unimodal validation data \u2014 text\n# Sources: dclm_val\n\nmodalities:\n- text\nval_steps: 200\nsampler_type: uniform\nsources:\n- name: dclm_val\n shards: data/shards/dclm_reshuffled/val/shard_{000000..000049}.tar\n weight: 50000\n modality_tokens:\n text: 1300\n format: npz\n"
|
| 109 |
+
}
|
| 110 |
+
}
|