INDEX

Explanations

references to pairs of objects or items

oai_token-act-pair · gpt-3.5-turbo

phrases relating to pairs or combinations of items or concepts

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

ulhu

-0.75

emetery

-0.73

schild

-0.72

inez

-0.70

INA

-0.67

ADRA

-0.67

 Occupations

-0.66

UGE

-0.64

abad

-0.64

 Causes

-0.63

POSITIVE LOGITS

ings

1.10

wise

1.00

lihood

0.96

pair

0.84

rings

0.82

horn

0.81

wich

0.80

 mates

0.78

 mate

0.74

hood

0.73

Activations Density 0.046%

references to pairs of objects or items

phrases relating to pairs or combinations of items or concepts

No Known Activations