INDEX

Explanations

references to or descriptions related to prominent or notable individuals

oai_token-act-pair · gpt-3.5-turbo

references to "figures" in various contexts, especially in visual or descriptive language

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

IFE

-0.74

termination

-0.73

DAQ

-0.72

ModLoader

-0.72

velength

-0.71

umenthal

-0.70

é¾

-0.69

bey

-0.69

ntil

-0.68

cyclopedia

-0.67

POSITIVE LOGITS

head

1.00

 prominently

1.00

 skating

0.99

heads

0.88

enance

0.88

books

0.77

 macros

0.75

figure

0.75

 figures

0.75

downs

0.73

Activations Density 0.036%

references to or descriptions related to prominent or notable individuals

references to "figures" in various contexts, especially in visual or descriptive language

No Comments

No Known Activations