INDEX

Explanations

Instances where the word "score" appears

oai_token-act-pair · gpt-3.5-turbo

references to scoring and points 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 @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 detachment

-0.68

 withdrawing

-0.62

 suitcase

-0.61

 Spice

-0.59

 mole

-0.59

agan

-0.59

 Parenthood

-0.57

 bust

-0.57

 Hyde

-0.57

 Boris

-0.56

POSITIVE LOGITS

card

1.43

cards

1.37

keeper

1.04

keepers

1.04

rs

1.02

stre

1.00

keeping

1.00

ific

0.99

heet

0.96

emi

0.86

Activations Density 0.038%

Instances where the word "score" appears

references to scoring and points in competitive contexts

No Comments

No Known Activations