Datasets:
Document dense-100m configuration
Browse files
README.md
CHANGED
|
@@ -19,6 +19,12 @@ configs:
|
|
| 19 |
path: "data/train/*"
|
| 20 |
- split: validation
|
| 21 |
path: "data/validation/*"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
---
|
| 23 |
|
| 24 |
<div align="center">
|
|
@@ -35,6 +41,10 @@ One deterministic corpus. Three training views: **pretraining · SFT · on-polic
|
|
| 35 |
|:--|:--|:--|:--|
|
| 36 |
| **10,000,000 rows** | **36 families · M1–M11** | **7.80 GiB · 41 Parquet shards** | **9,949,600 train · 50,400 validation** |
|
| 37 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 38 |
StateBench teaches models to execute finite state programs rather than recall facts. A typed program is executed by a deterministic reference evaluator; only then is it rendered into varied textual surfaces.
|
| 39 |
|
| 40 |
## Load it
|
|
@@ -42,11 +52,15 @@ StateBench teaches models to execute finite state programs rather than recall fa
|
|
| 42 |
from datasets import load_dataset
|
| 43 |
|
| 44 |
ds = load_dataset("aabbdev/StateBench-v1", split="train", streaming=True)
|
|
|
|
| 45 |
|
| 46 |
pretrain = ds.select_columns(["text"])
|
| 47 |
sft = ds.select_columns(["prompt", "completion"])
|
| 48 |
sdft = ds.select_columns(["prompt", "privileged_context"])
|
| 49 |
|
|
|
|
|
|
|
|
|
|
| 50 |
## One corpus, three views
|
| 51 |
|
| 52 |
| Regime | Student input | Training signal |
|
|
|
|
| 19 |
path: "data/train/*"
|
| 20 |
- split: validation
|
| 21 |
path: "data/validation/*"
|
| 22 |
+
- config_name: dense-100m
|
| 23 |
+
data_files:
|
| 24 |
+
- split: train
|
| 25 |
+
path: "data/dense-100m/train/*"
|
| 26 |
+
- split: validation
|
| 27 |
+
path: "data/dense-100m/validation/*"
|
| 28 |
---
|
| 29 |
|
| 30 |
<div align="center">
|
|
|
|
| 41 |
|:--|:--|:--|:--|
|
| 42 |
| **10,000,000 rows** | **36 families · M1–M11** | **7.80 GiB · 41 Parquet shards** | **9,949,600 train · 50,400 validation** |
|
| 43 |
|
| 44 |
+
| Dense extension | Physical rows | Density | Storage |
|
| 45 |
+
|:--|:--|:--|:--|
|
| 46 |
+
| **100,000,000 episodes** | **25,000,000 packed rows** | **162.01 units/1k chars** | **94.78 GiB · 251 shards** |
|
| 47 |
+
|
| 48 |
StateBench teaches models to execute finite state programs rather than recall facts. A typed program is executed by a deterministic reference evaluator; only then is it rendered into varied textual surfaces.
|
| 49 |
|
| 50 |
## Load it
|
|
|
|
| 52 |
from datasets import load_dataset
|
| 53 |
|
| 54 |
ds = load_dataset("aabbdev/StateBench-v1", split="train", streaming=True)
|
| 55 |
+
dense = load_dataset("aabbdev/StateBench-v1", "dense-100m", split="train", streaming=True)
|
| 56 |
|
| 57 |
pretrain = ds.select_columns(["text"])
|
| 58 |
sft = ds.select_columns(["prompt", "completion"])
|
| 59 |
sdft = ds.select_columns(["prompt", "privileged_context"])
|
| 60 |
|
| 61 |
+
The default configuration stores text directly. The dense-100m configuration omits that
|
| 62 |
+
redundant physical column; derive it from prompt and completion when using causal pretraining.
|
| 63 |
+
|
| 64 |
## One corpus, three views
|
| 65 |
|
| 66 |
| Regime | Student input | Training signal |
|