INDEX

Explanations

words related to cost, technology features, and specific product or company names

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

=-=-=-=-

-0.80

ename

-0.75

bats

-0.75

WAR

-0.74

ynt

-0.74

bright

-0.74

quet

-0.73

idelines

-0.73

said

-0.73

upper

-0.72

POSITIVE LOGITS

edly

0.88

 solutions

0.78

 affair

0.76

 earners

0.75

 versions

0.70

 thinker

0.69

TVs

0.69

 entertainment

0.68

 scenario

0.68

 telev

0.68

Activations Density 1.976%

words related to cost, technology features, and specific product or company names

No Comments

No Known Activations