INDEX

Explanations

phrases related to specific named entities with capitalized words comprising an acronym followed by a numerical designation

oai_token-act-pair · gpt-3.5-turbo

occurrences of the abbreviation "NE" 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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 slug

-0.68

 Crimson

-0.68

 hostage

-0.68

scr

-0.65

 Jackie

-0.64

 birth

-0.62

 depend

-0.61

 Ruby

-0.60

 baptism

-0.59

 incarn

-0.59

POSITIVE LOGITS

NE

4.17

NE

2.11

ne

1.79

NW

1.76

NES

1.58

NA

1.57

NEY

1.50

NO

1.50

NB

1.50

SE

1.49

Activations Density 0.007%

phrases related to specific named entities with capitalized words comprising an acronym followed by a numerical designation

occurrences of the abbreviation "NE" in various contexts

No Comments

No Known Activations