INDEX

Explanations

words related to leftovers or remaining parts of something

oai_token-act-pair · gpt-3.5-turbo

instances of the term "over" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

udes

-0.70

risome

-0.61

 commanded

-0.60

 attm

-0.59

ulet

-0.59

UES

-0.58

quo

-0.57

WI

-0.56

 guilty

-0.55

 whistle

-0.55

POSITIVE LOGITS

lord

0.93

ride

0.91

whelming

0.88

lords

0.84

rule

0.84

lay

0.83

flow

0.82

aturated

0.81

lap

0.79

leaf

0.78

Activations Density 0.009%

words related to leftovers or remaining parts of something

instances of the term "over" in various contexts

No Comments

No Known Activations