INDEX

Explanations

terms related to close combat or physical conflicts

oai_token-act-pair · gpt-3.5-turbo

references to combat and fighting terminology

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

oria

-0.86

love

-0.85

toc

-0.82

æ©

-0.82

redo

-0.80

gins

-0.80

gin

-0.79

opia

-0.76

ger

-0.74

 Lauder

-0.74

POSITIVE LOGITS

velength

0.92

 practition

0.89

 referen

0.85

nesday

0.85

enhagen

0.83

 councill

0.82

merce

0.79

aukee

0.78

 weaponry

0.75

yss

0.74

Activations Density 0.031%

terms related to close combat or physical conflicts

references to combat and fighting terminology

No Comments

No Known Activations