INDEX

Explanations

similes or comparisons using "like."

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

oust

-0.87

uty

-0.77

è¦ļéĨĴ

-0.75

Versions

-0.75

itles

-0.75

uid

-0.74

KA

-0.74

olphin

-0.73

inion

-0.72

arest

-0.71

POSITIVE LOGITS

lier

0.97

 something

0.93

 crap

0.93

 somebody

0.86

 someone

0.86

an

0.85

fun

0.83

 paradise

0.81

 cheating

0.79

Activations Density 0.057%

similes or comparisons using "like."

No Comments

No Known Activations