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

ighth

-0.80

irit

-0.77

phan

-0.76

 Cheong

-0.72

itsu

-0.71

repl

-0.70

otos

-0.70

yssey

-0.70

RESULTS

-0.70

-0.69

POSITIVE LOGITS

liest

0.72

 dispos

0.70

 ACTIONS

0.68

 ORIG

0.65

 Glob

0.64

 hats

0.64

 surn

0.64

geant

0.63

 Flan

0.63

 upfront

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.