INDEX

Explanations

mentions of the word "illicit" or variations thereof

oai_token-act-pair · gpt-3.5-turbo

words related to affiliation and belonging

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

 Wonderland

-0.65

 shining

-0.62

 place

-0.61

 royalties

-0.60

 clipping

-0.60

dex

-0.60

 Colorado

-0.59

 finalized

-0.59

 threshold

-0.58

 Dread

-0.58

POSITIVE LOGITS

ili

4.53

iliation

2.06

ilia

1.96

iliated

1.78

iliate

1.73

ilian

1.71

ilic

1.52

ile

1.51

iliary

1.51

illi

1.49

Activations Density 0.006%

mentions of the word "illicit" or variations thereof

words related to affiliation and belonging

No Comments

No Known Activations