INDEX

Explanations

names of individuals or entities in a list of text data

oai_token-act-pair · gpt-3.5-turbo

instances of the conjunction "and."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

advertisement

-0.70

esthetic

-0.66

etting

-0.64

beck

-0.64

eat

-0.63

ãĥīãĥ©

-0.62

ctory

-0.61

advertising

-0.61

":{"

-0.61

ective

-0.60

POSITIVE LOGITS

 others

1.67

 countless

1.30

 Others

1.18

 many

1.17

 assorted

1.15

 innumerable

1.11

 elsewhere

1.10

 dozens

1.08

 other

1.04

 numerous

1.02

Activations Density 0.168%

names of individuals or entities in a list of text data

instances of the conjunction "and."

No Comments

No Known Activations