INDEX

Explanations

words related to procrastination

oai_token-act-pair · gpt-3.5-turbo

variations of the word "rast", likely indicating a focus on specific actions or statuses related to something being completed

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

 Hayden

-0.85

TX

-0.71

 induced

-0.71

 unconditional

-0.67

 released

-0.66

 report

-0.66

 authorized

-0.65

 ready

-0.65

 exclusive

-0.64

 uncle

-0.63

POSITIVE LOGITS

rast

4.79

 Contrast

1.20

rals

1.18

ram

1.04

rarily

1.03

romy

1.01

rastructure

1.00

yrinth

1.00

rises

0.99

rad

0.98

Activations Density 0.012%

words related to procrastination

variations of the word "rast", likely indicating a focus on specific actions or statuses related to something being completed

No Comments

No Known Activations