INDEX

Explanations

forms of the word 'writer'

oai_token-act-pair · gpt-3.5-turbo

references to writers and writing

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

ADRA

-0.81

ibaba

-0.78

eneg

-0.76

ĸļ

-0.76

rals

-0.75

illon

-0.75

ypes

-0.68

 Lumpur

-0.68

 CCTV

-0.67

undai

-0.66

POSITIVE LOGITS

 laureate

0.96

writer

0.94

 writer

0.90

uscript

0.87

 fiction

0.86

writing

0.85

writers

0.81

writ

0.80

 Beware

0.76

ãĤ¼

0.75

Activations Density 0.022%

forms of the word 'writer'

references to writers and writing

No Comments

No Known Activations