Upload provenance/prep_stats.json with huggingface_hub
Browse files- provenance/prep_stats.json +94 -0
provenance/prep_stats.json
ADDED
|
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"seq_len_cap": 65536,
|
| 3 |
+
"train_rows": 211837,
|
| 4 |
+
"val_rows": 256,
|
| 5 |
+
"train_tokens_total": 2535132827,
|
| 6 |
+
"train_len_mean": 11967.4,
|
| 7 |
+
"train_len_p50": 6030,
|
| 8 |
+
"train_len_p90": 33309,
|
| 9 |
+
"train_len_p99": 59792,
|
| 10 |
+
"train_len_max": 65536,
|
| 11 |
+
"packing_sim": {
|
| 12 |
+
"32768": {
|
| 13 |
+
"bins": 53444,
|
| 14 |
+
"wasted_tokens": 783877001,
|
| 15 |
+
"leftover_tokens": 2834,
|
| 16 |
+
"steps_1ep": 3340,
|
| 17 |
+
"waste_pct": 30.92,
|
| 18 |
+
"tokens_per_step": 524288
|
| 19 |
+
},
|
| 20 |
+
"49152": {
|
| 21 |
+
"bins": 40038,
|
| 22 |
+
"wasted_tokens": 567182217,
|
| 23 |
+
"leftover_tokens": 2834,
|
| 24 |
+
"steps_1ep": 2502,
|
| 25 |
+
"waste_pct": 22.37,
|
| 26 |
+
"tokens_per_step": 786432
|
| 27 |
+
},
|
| 28 |
+
"65536": {
|
| 29 |
+
"bins": 31773,
|
| 30 |
+
"wasted_tokens": 452808294,
|
| 31 |
+
"leftover_tokens": 49205,
|
| 32 |
+
"steps_1ep": 3971,
|
| 33 |
+
"waste_pct": 17.86,
|
| 34 |
+
"tokens_per_step": 524288
|
| 35 |
+
}
|
| 36 |
+
},
|
| 37 |
+
"stats": {
|
| 38 |
+
"nemotron/raw": 204800,
|
| 39 |
+
"nemotron/non_thinking": 541045,
|
| 40 |
+
"nemotron/with_tool_calls": 1170572,
|
| 41 |
+
"nemotron/with_tools_defs": 82481,
|
| 42 |
+
"nemotron/thinking": 885735,
|
| 43 |
+
"benchmaxx/raw": 12397,
|
| 44 |
+
"benchmaxx/non_thinking": 12397,
|
| 45 |
+
"benchmaxx_cot/raw": 15877,
|
| 46 |
+
"benchmaxx_cot/non_thinking": 15877,
|
| 47 |
+
"nemotron/overlong_tokens_dropped": 596520073
|
| 48 |
+
},
|
| 49 |
+
"drops": {
|
| 50 |
+
"nemotron/bad_tools": 13542,
|
| 51 |
+
"nemotron/assistant_without_gen_prompt": 355,
|
| 52 |
+
"nemotron/think_tags_or_bad_tool_call": 21,
|
| 53 |
+
"nemotron/overlong_gt_65536": 7063
|
| 54 |
+
},
|
| 55 |
+
"nemotron_categories": {
|
| 56 |
+
"agentic": 48924,
|
| 57 |
+
"math": 28672,
|
| 58 |
+
"multilingual": 10240,
|
| 59 |
+
"code": 28672,
|
| 60 |
+
"science": 24576,
|
| 61 |
+
"math_proofs": 8192,
|
| 62 |
+
"arc_agi": 10240,
|
| 63 |
+
"search": 4096,
|
| 64 |
+
"swe": 16384,
|
| 65 |
+
"cuda": 2276,
|
| 66 |
+
"chat": 22528
|
| 67 |
+
},
|
| 68 |
+
"rescue_empty_string_tools": {
|
| 69 |
+
"affected": 13542,
|
| 70 |
+
"rescued": 13160,
|
| 71 |
+
"new_train_rows": 224997,
|
| 72 |
+
"new_train_tokens_total": 2592452318,
|
| 73 |
+
"packing_sim_65536": {
|
| 74 |
+
"bins": 32491,
|
| 75 |
+
"wasted_tokens": 463121778,
|
| 76 |
+
"leftover_tokens": 364,
|
| 77 |
+
"steps_1ep": 4061,
|
| 78 |
+
"waste_pct": 17.86
|
| 79 |
+
}
|
| 80 |
+
},
|
| 81 |
+
"monster_filter": {
|
| 82 |
+
"train_kept": 224991,
|
| 83 |
+
"train_dropped": 6,
|
| 84 |
+
"val_kept": 256,
|
| 85 |
+
"val_dropped": 0,
|
| 86 |
+
"train_tokens_total": 2592085287,
|
| 87 |
+
"packing_sim_65536": {
|
| 88 |
+
"bins": 32547,
|
| 89 |
+
"wasted_tokens": 459066386,
|
| 90 |
+
"leftover_tokens": 18709,
|
| 91 |
+
"steps_1ep": 4068
|
| 92 |
+
}
|
| 93 |
+
}
|
| 94 |
+
}
|