INDEX

Explanations

expressions related to instructions, demands, and warnings

oai_token-act-pair · gpt-3.5-turbo

actions associated with warnings, demands, and calls for accountability

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

AND

-0.66

and

-0.62

irement

-0.60

izza

-0.58

resents

-0.54

elta

-0.53

ÃŃa

-0.53

ectar

-0.52

brow

-0.52

gage

-0.52

POSITIVE LOGITS

 accordingly

1.47

 consequently

0.99

 therefore

0.93

 thereafter

0.93

 thereof

0.92

 consequ

0.85

 therein

0.85

 others

0.83

 subsequent

0.82

 thereto

0.82

Activations Density 0.533%

expressions related to instructions, demands, and warnings

actions associated with warnings, demands, and calls for accountability

No Comments

No Known Activations