INDEX

Explanations

proper nouns

oai_token-act-pair · gpt-3.5-turbo

instances of specific names or topics related to notable individuals or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 fall

-0.73

 brood

-0.67

 simulator

-0.64

 humour

-0.62

Graph

-0.61

 animated

-0.59

 shorthand

-0.59

 Falling

-0.59

 static

-0.57

 source

-0.56

POSITIVE LOGITS

tta

4.78

tti

2.34

tto

2.10

ttes

1.75

tt

1.63

lla

1.35

lli

1.24

ta

1.22

zza

1.20

lda

1.20

Activations Density 0.013%

proper nouns

instances of specific names or topics related to notable individuals or entities

No Comments

No Known Activations