INDEX

Explanations

information or descriptions that are detailed

oai_token-act-pair · gpt-3.5-turbo

instances of the word "detailed" indicating thoroughness or specificity

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

aden

-0.81

enos

-0.78

eno

-0.75

qqa

-0.73

ï¸

-0.70

lua

-0.69

Wee

-0.68

oi

-0.68

avis

-0.68

ADS

-0.68

POSITIVE LOGITS

 detail

1.04

 details

0.95

 descriptions

0.92

lly

0.83

 description

0.82

 representations

0.80

 outline

0.79

 examination

0.79

 detailing

0.79

 particulars

0.78

Activations Density 0.014%

information or descriptions that are detailed

instances of the word "detailed" indicating thoroughness or specificity

No Comments

No Known Activations