INDEX

Explanations

phrases related to unspecified activities or events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "some."

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

enance

-0.76

SPONSORED

-0.75

ettes

-0.72

rollers

-0.70

vest

-0.68

thus

-0.67

meet

-0.66

gon

-0.65

rors

-0.63

agents

-0.63

POSITIVE LOGITS

place

0.95

how

0.91

ones

0.87

body

0.84

 degree

0.81

etheless

0.76

 consolation

0.76

 semblance

0.72

HOW

0.70

 additional

0.68

Activations Density 0.083%

phrases related to unspecified activities or events

instances of the word "some."

No Comments

No Known Activations