INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Vale

-0.70

 Shepherd

-0.68

Til

-0.64

Ou

-0.63

 speculate

-0.63

rology

-0.62

 Ramsay

-0.62

REL

-0.61

 Gunn

-0.61

 Lastly

-0.60

POSITIVE LOGITS

flex

0.77

 wrist

0.74

paren

0.71

imar

0.69

bley

0.68

enger

0.65

bands

0.65

leigh

0.65

SPA

0.64

igmatic

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.