Commit
·
5cad637
1
Parent(s):
25416cc
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
@@ -1,55 +1,13 @@
|
|
1 |
-
|
2 |
-
dataset_info:
|
3 |
-
features:
|
4 |
-
- name: layer
|
5 |
-
dtype: uint8
|
6 |
-
- name: neuron
|
7 |
-
dtype: uint16
|
8 |
-
- name: rank
|
9 |
-
dtype: uint8
|
10 |
-
- name: activation
|
11 |
-
dtype: float32
|
12 |
-
- name: position
|
13 |
-
dtype: uint8
|
14 |
-
- name: id0
|
15 |
-
dtype: uint16
|
16 |
-
- name: id1
|
17 |
-
dtype: uint16
|
18 |
-
- name: id2
|
19 |
-
dtype: uint16
|
20 |
-
- name: id3
|
21 |
-
dtype: uint16
|
22 |
-
- name: id4
|
23 |
-
dtype: uint16
|
24 |
-
- name: id5
|
25 |
-
dtype: uint16
|
26 |
-
- name: id6
|
27 |
-
dtype: uint16
|
28 |
-
- name: id7
|
29 |
-
dtype: uint16
|
30 |
-
- name: id8
|
31 |
-
dtype: uint16
|
32 |
-
- name: id9
|
33 |
-
dtype: uint16
|
34 |
-
- name: id10
|
35 |
-
dtype: uint16
|
36 |
-
- name: id11
|
37 |
-
dtype: uint16
|
38 |
-
- name: id12
|
39 |
-
dtype: uint16
|
40 |
-
- name: id13
|
41 |
-
dtype: uint16
|
42 |
-
- name: id14
|
43 |
-
dtype: uint16
|
44 |
-
- name: id15
|
45 |
-
dtype: uint16
|
46 |
-
splits:
|
47 |
-
- name: train
|
48 |
-
num_bytes: 1934622720
|
49 |
-
num_examples: 47185920
|
50 |
-
download_size: 3184868008
|
51 |
-
dataset_size: 1934622720
|
52 |
-
---
|
53 |
-
# Dataset Card for "pythia-12b-neuron-dataset-examples"
|
54 |
|
55 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# pythia-12b-neuron-dataset-examples
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
+
This dataset contains the top 64 highest activating dataset examples for each
|
4 |
+
MLP neuron in Pythia-12b. The dataset examples are all 16 tokens long. See
|
5 |
+
https://confirmlabs.org/posts/dreaming.html for details.
|
6 |
+
|
7 |
+
Columns:
|
8 |
+
- `layer`: the layer of the neuron
|
9 |
+
- `neuron`: the index of the neuron
|
10 |
+
- `rank`: the rank of the example
|
11 |
+
- `activation`: the activation of the neuron on the example
|
12 |
+
- `position`: the token position for which the neuron is maximally activated.
|
13 |
+
- `id#`: the token id at position `#` in 0 to 15.
|