INDEX

Explanations

negative statements including the word "no."

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a negation or absence of something

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

eals

-0.77

ributes

-0.77

ean

-0.76

 Actions

-0.75

Js

-0.73

rs

-0.72

items

-0.71

mates

-0.71

ULTS

-0.70

anners

-0.70

POSITIVE LOGITS

 shortage

1.43

 doubt

1.34

 indication

1.18

 denying

1.16

 guarantee

1.13

 reason

1.11

 excuse

1.07

 ambiguity

1.06

 conceivable

1.04

 escaping

1.03

Activations Density 0.076%

negative statements including the word "no."

phrases indicating a negation or absence of something

No Comments

No Known Activations