INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 opioids

-0.73

 Pathfinder

-0.71

 Licensed

-0.70

 pills

-0.69

 Premium

-0.69

DRM

-0.68

ppm

-0.66

PST

-0.66

 genetically

-0.66

ãĤ¡

-0.65

POSITIVE LOGITS

ipedia

0.88

haw

0.80

verbs

0.80

swer

0.78

rey

0.76

sung

0.76

tip

0.75

ijn

0.75

Ĥ¬

0.75

adh

0.75

Activations Density 0.000%

No Known Activations

This feature has no known activations.