INDEX

Explanations

phrases that mention physical or digital locks

oai_token-act-pair · gpt-3.5-turbo

instances of the word "lock" in various contexts

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

ANN

-0.61

 orally

-0.59

oun

-0.58

hetical

-0.58

 appreci

-0.58

 inacc

-0.57

anqu

-0.56

 intervening

-0.56

issan

-0.55

 cognition

-0.54

POSITIVE LOGITS

lock

1.22

locking

1.00

heed

0.97

lear

0.92

locks

0.91

picking

0.88

er

0.86

eries

0.84

trap

0.81

itus

0.79

Activations Density 0.006%

phrases that mention physical or digital locks

instances of the word "lock" in various contexts

No Comments

No Known Activations