INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

found

-0.92

haven

-0.81

duct

-0.79

ldon

-0.77

iHUD

-0.76

sonian

-0.76

tex

-0.75

cit

-0.71

netflix

-0.70

article

-0.69

POSITIVE LOGITS

 typing

0.64

 memor

0.63

 badges

0.62

 verification

0.61

 reporting

0.61

 verbs

0.60

 Friendship

0.59

 courage

0.58

 metic

0.58

 copying

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.