INDEX

Explanations

scientific figures and references in a document

oai_token-act-pair · gpt-3.5-turbo

references to figures or illustrations in the text

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

 academy

-0.73

 booze

-0.70

 stuff

-0.70

 oath

-0.68

 perks

-0.68

civ

-0.67

 specials

-0.67

 reserved

-0.67

 consent

-0.67

 politics

-0.66

POSITIVE LOGITS

Figure

4.03

 Figure

3.08

Fig

2.16

 Figures

2.13

FIG

2.02

figure

1.93

FIG

1.72

Table

1.54

Fig

1.47

 figure

1.41

Activations Density 0.023%

scientific figures and references in a document

references to figures or illustrations in the text

No Comments

No Known Activations