INDEX

Explanations

generic terms or concepts that can be broadly categorized in different fields

oai_token-act-pair · gpt-3.5-turbo

references to general purpose categories and terms that express intent or definition

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

ogun

-0.70

 embargo

-0.67

ĻĤ

-0.64

aps

-0.63

hips

-0.63

 behold

-0.63

aeus

-0.62

ħĭ

-0.62

 gust

-0.62

 oppressive

-0.61

POSITIVE LOGITS

purpose

1.31

theless

1.26

POSE

1.04

tenance

0.96

terday

0.94

REDACTED

0.90

OPLE

0.88

racuse

0.86

ãĤ±

0.83

nsic

0.81

Activations Density 0.009%

generic terms or concepts that can be broadly categorized in different fields

references to general purpose categories and terms that express intent or definition

No Comments

No Known Activations