INDEX

Explanations

words related to organized and possibly secretive operations or missions

oai_token-act-pair · gpt-3.5-turbo

mentions of operations, particularly in a context related to law enforcement or organized activities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

gets

-0.74

cknowled

-0.73

reath

-0.72

bitious

-0.69

solid

-0.64

misc

-0.62

html

-0.62

nature

-0.62

 Scores

-0.61

ende

-0.61

POSITIVE LOGITS

 operation

1.33

 operations

1.01

 operative

0.91

operation

0.90

eering

0.88

 Operation

0.84

Operation

0.81

 procedure

0.80

 operated

0.79

chwitz

0.78

Activations Density 0.011%

words related to organized and possibly secretive operations or missions

mentions of operations, particularly in a context related to law enforcement or organized activities

No Comments

No Known Activations