INDEX

Explanations

text related to news, journalism, and publications

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

etts

-0.73

imer

-0.66

 Hann

-0.65

abal

-0.65

geist

-0.64

ilts

-0.63

bons

-0.63

ndum

-0.62

illon

-0.61

âĢİ

-0.60

POSITIVE LOGITS

interstitial

0.83

iframe

0.81

 subscribing

0.72

Subscribe

0.69

 Kavanaugh

0.69

Content

0.68

{*

0.68

APD

0.67

osion

0.65

 domains

0.64

Activations Density 1.786%

text related to news, journalism, and publications

No Comments

No Known Activations