evaluate 0 0 0

kyubeen/temp

published by @kyubeen

Alignment dataset tracked in the /datasets sub-catalog.

Lifetime downloads
222
Last 30 days
23
Likes
-
Size
38.1 MB
Created on HF
2026-03-21
Age
7 months ago

Description

JJS Eval Results

Benchmark Results Summary

Benchmark

Metric

Score

GSM8K

Pass@1 (symbolic_correct)

80.97%

Hendrycks Math

Pass@1 (symbolic_correct)

63.78%

Minerva Math

Pass@1 (symbolic_correct)

25.74%

Human-Eval

Pass@1 (base / plus)

59.15% / 53.05%

MBPP

Pass@1 (base / plus)

68.78% / 55.82%

IFEval

Average Score

56.29%

IFBench

Average Score

15.96%

Detailed Results

Math Benchmarks

GSM8K

Hendrycks Math… See the full description on the dataset page: https://huggingface.co/datasets/kyubeen/temp.

Tags

region:us

README current version from Hugging Face

JJS Eval Results

Benchmark Results Summary

Benchmark Metric Score
GSM8K Pass@1 (symbolic_correct) 80.97%
Hendrycks Math Pass@1 (symbolic_correct) 63.78%
Minerva Math Pass@1 (symbolic_correct) 25.74%
Human-Eval Pass@1 (base / plus) 59.15% / 53.05%
MBPP Pass@1 (base / plus) 68.78% / 55.82%
IFEval Average Score 56.29%
IFBench Average Score 15.96%

Detailed Results

Math Benchmarks

GSM8K Hendrycks Math Minerva Math
Entries 1,319 5,000 272
Pass@1 80.97% 63.78% 25.74%
No Answer Rate 0.76% 4.58% 6.99%
Avg Tokens 310 608 566

Code Benchmarks

Human-Eval MBPP
Entries 164 378
Pass@1 (base) 59.15% 68.78%
Pass@1 (plus) 53.05% 55.82%
Avg Tokens 179 60

Instruction Following Benchmarks

IFEval IFBench
Prompts 541 294
Instructions 834 335
Prompt Strict Acc 49.91% 12.93%
Instruction Strict Acc 60.31% 15.52%
Instruction Loose Acc 62.83% 19.40%
Average Score 56.29% 15.96%

File Structure

Each benchmark directory contains:

  • metrics.json - Summary metrics
  • output.jsonl - Detailed generation outputs

README history 2 revisions

Every night we snapshot the README of every dataset in the catalog. When the SHA changes we archive the new version and diff it against the last. This is the evolving thought record of the alignment-data field: what the author decided to say about the corpus, and how that framing shifted over time.

  1. 2026-03-21Upload folder using huggingface_hubb5a0ddb1.4 KB
    Loading...
  2. 2026-03-21initial commit6b9cd7b0 B
    Loading...

Used by abliterated models none yet

No abliterated models in our catalog have declared this dataset in their YAML metadata yet. This can mean the dataset is used but not declared, is used outside abliteration workflows, or is new. The link will populate automatically as models are indexed.

Related datasets same stage · evaluate

Read further

Repository

Download the dataset

0

Copy any of these snippets into your notebook or terminal. All three fetch directly from Hugging Face using your own credentials.

Recommended: datasets library (Python)
from datasets import load_dataset
ds = load_dataset("kyubeen/temp")
Raw snapshot (Python)
from huggingface_hub import snapshot_download
snapshot_download(repo_id="kyubeen/temp", repo_type="dataset")
git clone (requires git-lfs)
git clone https://huggingface.co/datasets/kyubeen/temp
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration