INDEX

Explanations

words related to winning or being champions

oai_token-act-pair · gpt-3.5-turbo

terms related to championships and notable achievements

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

 Pilot

-0.72

ringe

-0.70

amorph

-0.68

 Hoffman

-0.67

release

-0.65

Oxy

-0.65

MAX

-0.65

 Trooper

-0.64

hoe

-0.64

 cowboy

-0.63

POSITIVE LOGITS

 champions

2.30

ardy

1.53

 Scarborough

1.06

 Roose

0.98

 flagged

0.89

 Boxing

0.88

 substitutes

0.88

 harb

0.87

ampions

0.86

 hosts

0.83

Activations Density 0.026%

words related to winning or being champions

terms related to championships and notable achievements

No Comments

No Known Activations