INDEX

Explanations

mentions of people's work experience and professional interactions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "worked" to indicate employment or involvement in various roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

holders

-0.72

now

-0.65

SPONSORED

-0.64

abuse

-0.63

alter

-0.63

venge

-0.63

allows

-0.63

credit

-0.62

angan

-0.60

react

-0.60

POSITIVE LOGITS

ĸļ

0.93

nesday

0.93

CLASSIFIED

0.85

 destro

0.84

 tremend

0.83

 extensively

0.81

swick

0.79

 unsuccessfully

0.78

hement

0.78

ONSORED

0.75

Activations Density 0.144%

mentions of people's work experience and professional interactions

instances of the word "worked" to indicate employment or involvement in various roles

No Comments

No Known Activations