INDEX

Explanations

articles ('a', 'an') followed by a strong activation word

oai_token-act-pair · gpt-3.5-turbo

the word "a" in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Jagu

-0.81

Contents

-0.81

Edit

-0.78

ãĤ¬

-0.78

Ingredients

-0.74

Features

-0.72

onto

-0.72

Ont

-0.71

chuk

-0.71

Catal

-0.70

POSITIVE LOGITS

lot

1.16

 couple

1.08

cknowled

1.05

few

1.04

 bunch

1.03

 handful

0.99

 huge

0.94

rouse

0.94

 combination

0.94

 typical

0.93

Activations Density 0.859%

articles ('a', 'an') followed by a strong activation word

the word "a" in various contexts

No Comments

No Known Activations