zimplex's picture
Upload verified HELM artifacts [2d0602a5e5934c3dbf76eb922bdd1ead]
64b60ee verified
Raw History Blame Contribute Delete
5.64 kB
{
"health_reporting": "Native aggregate intervals; the last interval can span nonfinite detection. These are not per-step diagnostics.",
"paired_evaluation_points": 68,
"runs": [
{
"actual_terminal_step": 19600,
"attention_operand_format": "e2m1",
"evaluation_points": 9,
"final_c4_ppl": null,
"final_wikitext_ppl": null,
"first_nonfinite_step": 19576,
"full_budget_completed": false,
"initialization": "scaled",
"last_finite_evaluation": {
"c4_ppl": 1283.73413189331,
"step": 18000,
"wikitext_ppl": 2968.7661825600626
},
"last_finite_saved_step": 19500,
"native_cell_key": "scaled_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f4e2m1b32",
"native_json": "cells.d/scaled_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f4e2m1b32.json",
"native_json_sha256": "6ef01b6eb73a5a58a9a7a58d674b9052b33f8c6f43d13ae68586854312a4312e",
"planned_steps": 61000,
"planned_tokens": 1998848000,
"run": "scaled4",
"seed": 1234,
"status": "nonfinite"
},
{
"actual_terminal_step": 37800,
"attention_operand_format": "e4m3",
"evaluation_points": 18,
"final_c4_ppl": null,
"final_wikitext_ppl": null,
"first_nonfinite_step": 37785,
"full_budget_completed": false,
"initialization": "scaled",
"last_finite_evaluation": {
"c4_ppl": 64.18778187722458,
"step": 36000,
"wikitext_ppl": 178.46032642706157
},
"last_finite_saved_step": 37500,
"native_cell_key": "scaled_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f8e4m3b32",
"native_json": "cells.d/scaled_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f8e4m3b32.json",
"native_json_sha256": "566d1325afbb580f0d8485054e6b56e95eec2b96f2a7393a4c3a6a7cb5bff177",
"planned_steps": 61000,
"planned_tokens": 1998848000,
"run": "scaled8",
"seed": 1234,
"status": "nonfinite"
},
{
"actual_terminal_step": 21300,
"attention_operand_format": null,
"evaluation_points": 10,
"final_c4_ppl": null,
"final_wikitext_ppl": null,
"first_nonfinite_step": 21249,
"full_budget_completed": false,
"initialization": "unit",
"last_finite_evaluation": {
"c4_ppl": 1971.1374237795667,
"step": 20000,
"wikitext_ppl": 3768.617090639044
},
"last_finite_saved_step": 21000,
"native_cell_key": "unit_bf16_d910_c1_lr0.001_acc2_s7_st61000_ev2000_corpc42b_perm_cos_w8",
"native_json": "cells.d/unit_bf16_d910_c1_lr0.001_acc2_s7_st61000_ev2000_corpc42b_perm_cos_w8.json",
"native_json_sha256": "b8560c64449f735ab4d712e6a4df7aca7910bb856ed44e22bf2bf75768d549eb",
"planned_steps": 61000,
"planned_tokens": 1998848000,
"run": "unit_bf16_s7",
"seed": 7,
"status": "nonfinite"
},
{
"actual_terminal_step": 61000,
"attention_operand_format": null,
"evaluation_points": 31,
"final_c4_ppl": 42.14531938222052,
"final_wikitext_ppl": 150.1752538901007,
"first_nonfinite_step": null,
"full_budget_completed": true,
"initialization": "unit",
"last_finite_evaluation": {
"c4_ppl": 42.14531938222052,
"step": 61000,
"wikitext_ppl": 150.1752538901007
},
"last_finite_saved_step": null,
"native_cell_key": "unit_bf16_d910_c1_lr0.001_acc2_s99_st61000_ev2000_corpc42b_perm_cos_w8",
"native_json": "cells.d/unit_bf16_d910_c1_lr0.001_acc2_s99_st61000_ev2000_corpc42b_perm_cos_w8.json",
"native_json_sha256": "d396be378862680c3c53f90e0d7b29f77fbf178d1364d5af364b9a8f91202428",
"planned_steps": 61000,
"planned_tokens": 1998848000,
"run": "unit_bf16_s99",
"seed": 99,
"status": "finished"
},
{
"actual_terminal_step": 100,
"attention_operand_format": "e2m1",
"evaluation_points": 0,
"final_c4_ppl": null,
"final_wikitext_ppl": null,
"first_nonfinite_step": 94,
"full_budget_completed": false,
"initialization": "unit",
"last_finite_evaluation": null,
"last_finite_saved_step": null,
"native_cell_key": "unit_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f4e2m1b32",
"native_json": "cells.d/unit_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f4e2m1b32.json",
"native_json_sha256": "c8024707c0bcc82ac8f0d43e78fbc458c4f096f1cbbb97b04a711d161980872a",
"planned_steps": 61000,
"planned_tokens": 1998848000,
"run": "unit4",
"seed": 1234,
"status": "nonfinite"
},
{
"actual_terminal_step": 500,
"attention_operand_format": "e4m3",
"evaluation_points": 0,
"final_c4_ppl": null,
"final_wikitext_ppl": null,
"first_nonfinite_step": 470,
"full_budget_completed": false,
"initialization": "unit",
"last_finite_evaluation": null,
"last_finite_saved_step": null,
"native_cell_key": "unit_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f8e4m3b32",
"native_json": "cells.d/unit_bf16_d910_c1_lr0.001_acc2_s1234_st61000_ev2000_corpc42b_perm_cos_w8_f8e4m3b32.json",
"native_json_sha256": "c593f530ac19449f8adb4464c72383b51048ab95f629f2af5a3edff8ee2a6cf8",
"planned_steps": 61000,
"planned_tokens": 1998848000,
"run": "unit8",
"seed": 1234,
"status": "nonfinite"
}
],
"schema": "helm.c42b.public-results.v1"
}