INDEX

Explanations

punctuation marks indicating emphasis or dramatic effect

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

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

Mob

-0.71

 Paso

-0.70

ciples

-0.67

 Slug

-0.66

 Dragons

-0.62

 Elys

-0.62

 Folder

-0.59

 Panda

-0.59

 Stard

-0.59

 Gard

-0.58

POSITIVE LOGITS

 gasp

0.92

perhaps

0.91

albeit

0.89

again

0.88

––

0.86

almost

0.83

conserv

0.78

quite

0.77

 surprise

0.76

 importantly

0.76

Activations Density 0.100%

punctuation marks indicating emphasis or dramatic effect

No Comments

No Known Activations