INDEX

Explanations

words related to work or productivity

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "a" or the indefinite article in various contexts

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

Contents

-0.82

evidence

-0.81

ographs

-0.77

 Amend

-0.73

achu

-0.69

coins

-0.69

etsk

-0.69

ittens

-0.68

marks

-0.67

ONEY

-0.66

POSITIVE LOGITS

 bunch

1.08

 handful

1.03

 couple

1.00

lot

0.97

few

0.94

 multitude

0.93

 plethora

0.92

 whopping

0.92

 single

0.91

 dozen

0.90

Activations Density 0.296%

words related to work or productivity

references to the concept of "a" or the indefinite article in various contexts

No Comments

No Known Activations