INDEX

Explanations

years or ranges of years encoded in a specific format

oai_token-act-pair · gpt-3.5-turbo

references to specific years or numerical data in a coded format

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

ores

-0.81

alling

-0.79

alls

-0.76

 rele

-0.74

 RELE

-0.74

 Lotus

-0.71

ore

-0.70

ALS

-0.70

 Rory

-0.69

Lot

-0.69

POSITIVE LOGITS

mia

1.03

inc

0.80

agascar

0.79

iment

0.79

mb

0.78

inki

0.78

isma

0.77

icon

0.77

zman

0.76

axis

0.76

Activations Density 0.442%

years or ranges of years encoded in a specific format

references to specific years or numerical data in a coded format

No Comments

No Known Activations