INDEX

Explanations

descriptive words and phrases indicating a clever or deceptive technique

oai_token-act-pair · gpt-3.5-turbo

terms related to tricks and techniques

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ensus

-0.76

ALE

-0.71

 Afric

-0.64

 Corpus

-0.64

angan

-0.64

ital

-0.62

Ê

-0.61

 Tennessee

-0.59

 Cultural

-0.59

 Housing

-0.59

POSITIVE LOGITS

 tricks

4.07

 trick

2.19

 stunts

1.68

 Trick

1.57

 maneuvers

1.52

 shenanigans

1.51

 techniques

1.50

 tactics

1.41

 gimm

1.41

 glitches

1.33

Activations Density 0.014%

descriptive words and phrases indicating a clever or deceptive technique

terms related to tricks and techniques

No Comments

No Known Activations