INDEX

Explanations

phrases or words related to data visualization, such as "charts" and "graphs"

oai_token-act-pair · gpt-3.5-turbo

references to charts or data visualizations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

owner

-0.72

rea

-0.68

non

-0.68

ully

-0.66

iam

-0.64

+(

-0.64

--+

-0.62

EG

-0.62

owment

-0.62

ess

-0.62

POSITIVE LOGITS

 charts

4.00

 chart

2.21

 graphs

2.14

 Chart

1.84

chart

1.81

 diagrams

1.72

Chart

1.71

 forecasts

1.46

 tables

1.37

 maps

1.35

Activations Density 0.011%

phrases or words related to data visualization, such as "charts" and "graphs"

references to charts or data visualizations

No Comments

No Known Activations