INDEX

Explanations

references to fun and enjoyment in various contexts

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

©¶æ

-0.69

 Centauri

-0.68

 underest

-0.64

 mater

-0.62

 defective

-0.62

 condemned

-0.62

 adamant

-0.62

 defect

-0.61

 prophets

-0.61

inx

-0.61

POSITIVE LOGITS

nels

1.38

nel

1.02

gal

0.96

tainment

0.94

 Flavoring

0.92

icular

0.92

nell

0.88

cffff

0.86

atorium

0.85

imation

0.83

Activations Density 1.122%

references to fun and enjoyment in various contexts

No Comments

No Known Activations