INDEX

Explanations

names or terms with a particular pattern of letters or syllables

oai_token-act-pair · gpt-3.5-turbo

frequently occurring abbreviations, acronyms, or shorthand notations

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

è¦ļéĨĴ

-0.78

èĢħ

-0.74

 eleph

-0.73

govtrack

-0.72

ħĭ

-0.72

Ranked

-0.71

 DRAG

-0.70

Ļ½

-0.70

ãĥ¼ãĥĨ

-0.69

querque

-0.68

POSITIVE LOGITS

lich

0.89

ewater

0.85

itzer

0.79

kered

0.79

emaker

0.77

atever

0.77

nit

0.75

eworks

0.75

ework

0.74

bley

0.74

Activations Density 0.068%

names or terms with a particular pattern of letters or syllables

frequently occurring abbreviations, acronyms, or shorthand notations

No Comments

No Known Activations