INDEX

Explanations

text related to end-of-life discussions and encryption

oai_token-act-pair · gpt-3.5-turbo

negative phrases or concepts related to end-of-life issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ï¸ı

-0.65

usercontent

-0.61

aghan

-0.59

 Scream

-0.58

itsch

-0.57

 Protective

-0.56

BM

-0.56

 dracon

-0.55

HAEL

-0.55

 personally

-0.54

POSITIVE LOGITS

date

0.88

sale

0.83

around

0.82

life

0.82

anger

0.81

goal

0.81

users

0.81

angers

0.80

game

0.80

endment

0.80

Activations Density 0.037%

text related to end-of-life discussions and encryption

negative phrases or concepts related to end-of-life issues

No Comments

No Known Activations