INDEX

Explanations

references to the word "fluff"

oai_token-act-pair · gpt-3.5-turbo

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

CHR

-0.76

chrome

-0.65

Hunt

-0.65

 lapse

-0.64

xual

-0.64

 Cambod

-0.62

metic

-0.62

erg

-0.61

 freeze

-0.61

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

-0.60

POSITIVE LOGITS

ington

1.13

ield

1.04

IELD

0.99

iculty

0.91

uff

0.89

aneous

0.84

alo

0.82

ixel

0.82

ieri

0.82

ucket

0.81

Activations Density 0.015%

references to the word "fluff"

No Comments

No Known Activations