INDEX

Explanations

the word "winner" or phrases related to winning

oai_token-act-pair · gpt-3.5-turbo

mentions of "winner" and associated concepts in competitive contexts

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

rued

-0.65

Monitor

-0.62

 Anon

-0.60

ania

-0.60

ular

-0.60

express

-0.59

ements

-0.59

olic

-0.59

Dialogue

-0.59

allel

-0.58

POSITIVE LOGITS

 winner

3.88

 winners

2.83

winner

2.70

Winner

2.51

 Winner

2.47

 loser

2.12

 vict

2.00

 Winners

1.88

 winning

1.74

winning

1.67

Activations Density 0.021%

the word "winner" or phrases related to winning

mentions of "winner" and associated concepts in competitive contexts

No Comments

No Known Activations