INDEX

Explanations

contractions of specific phrases like "don't" and variations where "t" is used instead of an apostrophe

oai_token-act-pair · gpt-3.5-turbo

negative contractions indicating rejection or refusal

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

accompan

-0.69

ocative

-0.66

 Darling

-0.61

ricular

-0.58

forms

-0.57

rising

-0.57

anni

-0.56

leck

-0.55

itaire

-0.55

ò

-0.54

POSITIVE LOGITS

 ourselves

1.38

 know

0.91

ird

0.91

 condone

0.91

 underestimate

0.87

 need

0.86

 expect

0.84

bsite

0.81

asel

0.80

 hear

0.79

Activations Density 0.104%

contractions of specific phrases like "don't" and variations where "t" is used instead of an apostrophe

negative contractions indicating rejection or refusal

No Comments

No Known Activations