INDEX

Explanations

images or captions related to news articles and reports

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

aptic

-0.81

laws

-0.80

secut

-0.78

kson

-0.78

ishops

-0.75

unin

-0.75

idential

-0.72

ensable

-0.72

itte

-0.72

hurst

-0.71

POSITIVE LOGITS

 caption

1.12

 galleries

0.95

 gallery

0.92

gallery

0.88

img

0.85

 Gallery

0.85

 image

0.83

 macros

0.83

 depicting

0.83

 images

0.81

Activations Density 0.499%

images or captions related to news articles and reports

No Comments

No Known Activations