INDEX

Explanations

the prefix "pred-" or "predicated" followed by a concept or condition

oai_token-act-pair · gpt-3.5-turbo

mentions of prediction or premonition concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ierrez

-0.81

IFT

-0.76

ASE

-0.69

 Soup

-0.67

ãĥīãĥ©ãĤ´ãĥ³

-0.66

BOX

-0.63

Hub

-0.62

 Shake

-0.61

hiba

-0.61

OPLE

-0.61

POSITIVE LOGITS

efined

1.17

icip

1.03

icated

1.01

nis

1.01

icates

0.97

acent

0.94

 pred

0.92

etermin

0.91

ications

0.89

ication

0.88

Activations Density 0.006%

the prefix "pred-" or "predicated" followed by a concept or condition

mentions of prediction or premonition concepts

No Comments

No Known Activations