INDEX

Explanations

phrases indicating events or decisions that have not happened yet

oai_token-act-pair · gpt-3.5-turbo

phrases indicating something that has not happened or been decided yet

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

tein

-0.75

rities

-0.68

ricular

-0.67

ĪĴ

-0.66

itialized

-0.64

desktop

-0.63

 ashes

-0.62

sav

-0.62

Myth

-0.62

writers

-0.61

POSITIVE LOGITS

yet

0.76

hin

0.70

heric

0.70

 Brees

0.69

agher

0.68

ijn

0.67

 somehow

0.66

âĢ¦âĢ¦

0.65

terday

0.65

 lacking

0.64

Activations Density 0.023%

phrases indicating events or decisions that have not happened yet

phrases indicating something that has not happened or been decided yet

No Comments

No Known Activations