INDEX

Explanations

contractions of "were not" or "had not"

oai_token-act-pair · gpt-3.5-turbo

negative contractions related to existence or presence

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

[+]

-0.73

 DRAGON

-0.72

ONS

-0.68

 Mechdragon

-0.67

upp

-0.67

lees

-0.63

comings

-0.63

PU

-0.63

ramid

-0.62

 Dise

-0.61

POSITIVE LOGITS

't

1.07

itial

0.90

iting

0.73

gery

0.71

announced

0.70

ited

0.69

 tink

0.69

apolog

0.67

ajor

0.67

ÃŃ

0.67

Activations Density 0.049%

contractions of "were not" or "had not"

negative contractions related to existence or presence

No Comments

No Known Activations