INDEX

Explanations

phrases indicating a previous statement or mention

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "I said" or similar references to prior statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

quartered

-0.88

generated

-0.77

earances

-0.72

ococ

-0.70

BY

-0.70

asm

-0.70

mens

-0.69

ctors

-0.66

 Lauder

-0.64

bers

-0.64

POSITIVE LOGITS

 earlier

1.62

 above

1.34

 previously

1.25

 before

1.16

 previous

1.12

 yesterday

1.10

 BEFORE

1.00

 preceding

0.99

before

0.98

 beforehand

0.95

Activations Density 0.122%

phrases indicating a previous statement or mention

instances of the phrase "I said" or similar references to prior statements

No Comments

No Known Activations