INDEX

Explanations

the word "Lug" along with various other similar words with different activation strengths

oai_token-act-pair · gpt-3.5-turbo

references to specific locations and languages

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

-0.72

GAME

-0.69

 STATS

-0.68

éĹ

-0.67

 concess

-0.66

DC

-0.65

yson

-0.65

kins

-0.64

Jed

-0.64

CEPT

-0.63

POSITIVE LOGITS

Lug

0.97

rador

0.96

ansk

0.94

rats

0.92

enger

0.90

uing

0.90

roup

0.87

ues

0.86

emort

0.85

er

0.85

Activations Density 0.016%

the word "Lug" along with various other similar words with different activation strengths

references to specific locations and languages

No Comments

No Known Activations