INDEX

Explanations

words related to requests or instructions

oai_token-act-pair · gpt-3.5-turbo

references to specific items or concepts collectively identified as "these."

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

bow

-0.64

NESS

-0.55

IDA

-0.54

lement

-0.54

ONS

-0.53

ON

-0.52

ounge

-0.52

VIDEO

-0.51

ONE

-0.51

irez

-0.51

POSITIVE LOGITS

 these

2.71

these

2.35

 THESE

2.05

These

1.81

 These

1.66

 those

1.49

 this

1.42

those

1.28

 them

1.16

this

1.10

Activations Density 0.061%

words related to requests or instructions

references to specific items or concepts collectively identified as "these."

No Comments

No Known Activations