INDEX

Explanations

references to locations or spatial relations, specifically the word "across"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "the."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

SPONSORED

-0.74

uci

-0.72

olkien

-0.72

wine

-0.68

wu

-0.68

ATURES

-0.64

ata

-0.63

âĢł

-0.63

Ã©s

-0.63

antes

-0.63

POSITIVE LOGITS

 same

1.25

 latter

1.09

 entire

1.07

 remainder

1.04

 entirety

1.02

 ensuing

1.01

 aforementioned

1.00

 whole

1.00

 proverbial

0.99

 rest

0.99

Activations Density 0.827%

references to locations or spatial relations, specifically the word "across"

instances of the word "the."

No Comments

No Known Activations