INDEX

Explanations

specific strings of text related to player positions in football teams

oai_token-act-pair · gpt-3.5-turbo

references to specific football positions or player categories

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

asia

-0.82

ement

-0.81

cephal

-0.68

Plot

-0.68

 mortar

-0.67

tsy

-0.67

ãĤ¡

-0.66

Bas

-0.65

atra

-0.65

ãĥ³

-0.63

POSITIVE LOGITS

WR

3.82

WR

2.47

RB

1.80

TE

1.71

QB

1.53

CB

1.32

 receivers

1.32

AW

1.30

LB

1.26

KR

1.26

Activations Density 0.014%

specific strings of text related to player positions in football teams

references to specific football positions or player categories

No Comments

No Known Activations