INDEX

Explanations

the name "Barrett" with varying levels of activation

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals, particularly those named "Barrett" and "Kroger."

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

iverpool

-0.80

psons

-0.80

tein

-0.80

unciation

-0.75

uring

-0.74

uation

-0.72

ricular

-0.71

mercial

-0.69

imates

-0.69

iating

-0.68

POSITIVE LOGITS

 Barrett

0.89

ophon

0.78

 Bore

0.73

xon

0.72

ozy

0.70

lin

0.69

ãģĨ

0.68

olini

0.65

 Pharma

0.64

ĸļ

0.64

Activations Density 0.027%

the name "Barrett" with varying levels of activation

references to specific individuals, particularly those named "Barrett" and "Kroger."

No Comments

No Known Activations