INDEX

Explanations

phrases indicating high levels of readiness or activity

oai_token-act-pair · gpt-3.5-turbo

instances of the word "up."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 understatement

-0.69

 Discovery

-0.65

VICE

-0.64

 CLSID

-0.64

 hybrids

-0.62

 Frag

-0.61

awaru

-0.58

 Sorceress

-0.57

 compartment

-0.55

mology

-0.55

POSITIVE LOGITS

ended

1.18

raised

1.13

stairs

1.07

skirts

1.04

bra

0.99

ending

0.97

vote

0.97

regulated

0.95

standing

0.94

rated

0.93

Activations Density 0.054%

phrases indicating high levels of readiness or activity

instances of the word "up."

No Comments

No Known Activations