Vincent Fiszbin commited on
Commit ·
c23cd22
1
Parent(s): 4a50283
Add quantization calibration dataset
Browse files- README.md +27 -0
- data/train-00000-of-00001.parquet +3 -0
README.md
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
dataset_info:
|
| 3 |
+
features:
|
| 4 |
+
- name: question
|
| 5 |
+
dtype: string
|
| 6 |
+
- name: support
|
| 7 |
+
dtype: string
|
| 8 |
+
- name: choices
|
| 9 |
+
sequence: string
|
| 10 |
+
- name: answer
|
| 11 |
+
dtype: string
|
| 12 |
+
- name: source
|
| 13 |
+
dtype: string
|
| 14 |
+
splits:
|
| 15 |
+
- name: train
|
| 16 |
+
num_bytes: 176196
|
| 17 |
+
num_examples: 512
|
| 18 |
+
dataset_size: 176196
|
| 19 |
+
configs:
|
| 20 |
+
- config_name: default
|
| 21 |
+
data_files:
|
| 22 |
+
- split: train
|
| 23 |
+
path: data/train-*
|
| 24 |
+
---
|
| 25 |
+
This dataset contains a subset of the SCiQ multiple-choice science questions.
|
| 26 |
+
- Purpose: Calibration for weight and activation quantization of Qwen3-0.6B-Base
|
| 27 |
+
- Selection: 512 random samples from the SCiQ train set
|
data/train-00000-of-00001.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6358bf607203817a3ba85f5d0f7dea53eed4c35f46515d4227f7b1a2c19ae386
|
| 3 |
+
size 176196
|