INDEX

Explanations

words related to incorrectness or errors

oai_token-act-pair · gpt-3.5-turbo

instances of the word "wrong" and its variations, indicating errors or failures

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

enance

-0.73

ILA

-0.72

Ri

-0.67

 Swim

-0.66

 Flavoring

-0.64

tsky

-0.62

 Chill

-0.60

 CARE

-0.60

 Colbert

-0.59

kamp

-0.59

POSITIVE LOGITS

headed

1.40

fully

1.39

doing

1.04

do

1.03

fulness

0.98

ful

0.91

footed

0.90

sight

0.89

behavior

0.89

dest

0.89

Activations Density 0.029%

words related to incorrectness or errors

instances of the word "wrong" and its variations, indicating errors or failures

No Comments

No Known Activations