Datasets:
messages listlengths 2 2 | answer stringclasses 6
values | bin stringclasses 4
values | token_count int64 72 8.51k | original_idx int64 | target_entries int64 | object stringclasses 0
values |
|---|---|---|---|---|---|---|
[
{
"role": "user",
"content": "The \"incitements\" gave him courage, so that he exclaims: \"Shall I be of\nso untoward a disposition, as to refuse to lead the blind into the right\nway? Shall I be so unnatural, as not to give bread to the hungrie, or\nuncharitable, as not to cover the naked?\" It wasn't to b... | bathroom | 2k | 1,748 | null | null | null |
[
{
"role": "user",
"content": "Mary journeyed to the kitchen. John took the milk there. Daniel went to the garden. John discarded the milk. John went to the kitchen. Sandra moved to the bathroom. John moved to the garden. Mary moved to the garden. Mary went back to the bedroom. Daniel went to the bathroom. S... | office | 0k | 398 | null | null | null |
[
{
"role": "user",
"content": "\"Of course you're not,\" Pauline told her. \"It strikes me, for\ncompany--you're making yourself very much at home! She led the way along the hall to the spare room,\nthrowing the door wide open. Hilary cried, then stood quite still on the threshold, looking\nabout her with wi... | hallway | 4k | 3,694 | null | null | null |
[
{
"role": "user",
"content": "Nelson was slightly taken back by the answer; then an amused look came\ninto his face, and he said, in a bantering tone: \"Liked it, didn't you?\" \"I was just\nmad enough at you to kill you.\" \"There is the boy for me,\" said Nelson, turning to his staff. \"He not\nonly captu... | bathroom | 4k | 3,689 | null | null | null |
[
{
"role": "user",
"content": "The child’s heart has\ngrown suddenly light. The terrible fear which has been weighing her\ndown for the last few hours has been lifted, and Ruby is her old joyous\nself again. “Dad,” the little girl says later on. They are sitting out on the\nverandah, enjoying the comparative... | hallway | 4k | 3,132 | null | null | null |
[
{
"role": "user",
"content": "Daniel got the milk. Be that as it may, the fact remains, that throughout winter almost\nevery week's chronicle at the Lizard is the same story--wild storms, or\ndense fogs, guns of distress heard, a hasty manning of the life-boat,\ndragged with difficulty down the steep cliff-... | bedroom | 4k | 3,716 | null | null | null |
[
{
"role": "user",
"content": "When the eventful day arrived, the hands, instead of working till one,\nwere paid at twelve o'clock and rushed off home to have a wash and\nchange. The brakes were to start from the 'Cricketers' at one, but it was\narranged, for the convenience of those who lived at Windley, th... | bedroom | 8k | 7,570 | null | null | null |
[
{
"role": "user",
"content": "2 also shews a powerful mode of defending a field work by means of\nRockets, in addition to the defences of the present system; merely by\ncutting embrasures in the glacis, for horizontal firing. [Illustration: _Plate 7_ Fig. 1 Fig. 2]\n\n\n\n\nA ROCKET AMBUSCADE. 1, represen... | hallway | 4k | 3,708 | null | null | null |
[
{
"role": "user",
"content": "John took the milk. Purgatives should be repeated until all vicious humors are\ndischarged. Sydenham his descriptions of the epidemic which he witnessed in\nLondon in 1669-72 with the artistic touches of the master's hand. \"The\ndisease sets in,\" he says, \"with chills and s... | hallway | 4k | 3,780 | null | null | null |
[
{
"role": "user",
"content": "The carriage rolled briskly through the long streets into the smoky\ndown-town district. It arrived at the Auditorium, and Jennie was\nescorted to Lester's room. Daniel journeyed to the bedroom. He had talked\nlittle, leaving her to her thoughts. In this great hotel she felt\nd... | kitchen | 4k | 3,726 | null | null | null |
[
{
"role": "user",
"content": "He welcomed the first solitary row of darkened houses with a faint cheer\nof recognition. The scattered lamp-posts lightened his spirits, and even\nthe badly paved streets rang under the beats of his horse's feet like\nmusic. Great mills and manufactories, with only a night-wat... | bathroom | 4k | 3,737 | null | null | null |
[
{
"role": "user",
"content": "The next mare to make four figures at a\nstud sale was Hendre Crown Princess at the Lockinge sale of February\n14, 1900, the successful bidder being Mr. H. H. Smith-Carington,\nAshby Folville, Melton Mowbray, who has bought and bred many good\nShires. This date, February 14, se... | bathroom | 8k | 7,578 | null | null | null |
End of preview. Expand in Data Studio
Randomized YaRN — BABILong (QA3)
Frozen train/eval splits for the BABILong QA3 experiments in Randomized YaRN Improves Length Generalization for Long-Context Reasoning (Mehta, Yin, Durrett). Published so the paper's numbers reproduce independently of library/tokenizer versions — load these fixed bytes instead of regenerating.
Splits
| split | rows | description |
|---|---|---|
train |
4000 | 1000 subsampled QA3 examples per bin (0k/2k/4k/8k), seed 42 |
test |
2703 | the paper's multi-entry eval subset (305/bin; 264 at 1k, 304 at 2k) over bins 0k–128k |
Columns: messages (chat turns), answer, bin, token_count; test adds
original_idx, target_entries, object.
Provenance
Derived from RMT-team/babilong-train-5k-samples (train) and RMT-team/babilong-1k-samples
(eval), task qa3, both Apache-2.0. The multi-entry eval is filtered by a fixed 305-index
manifest. Generation: seed 42; subsample 1000/bin before tokenization. The default config's
two splits serve both Qwen and Olmo (the data is tokenizer-independent at rest).
- Downloads last month
- 22