INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

irlf

-0.70

bm

-0.70

 suspic

-0.69

 Bash

-0.69

 Happ

-0.68

dom

-0.65

 fundament

-0.62

 Barbarian

-0.62

 è£ıè

-0.61

 whereby

-0.61

POSITIVE LOGITS

ateur

0.78

 ultraviolet

0.75

%%%%

0.73

 acrylic

0.71

azeera

0.70

uds

0.67

 turkey

0.66

hire

0.66

0.63

UGH

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.