INDEX

Explanations

instances of things being related or connected to other things or actions

oai_token-act-pair · gpt-3.5-turbo

the word "as" indicating comparisons or similarities

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

anos

-0.89

xit

-0.77

wrong

-0.76

file

-0.75

utor

-0.72

orius

-0.71

CAR

-0.70

ambers

-0.70

amon

-0.69

mid

-0.69

POSITIVE LOGITS

 assorted

0.84

 ensuring

0.80

 optionally

0.79

 others

0.79

 those

0.77

 possibly

0.75

 comprising

0.74

 being

0.71

 providing

0.71

 preventing

0.70

Activations Density 0.048%

instances of things being related or connected to other things or actions

the word "as" indicating comparisons or similarities

No Comments

No Known Activations