INDEX

Explanations

capitalized words

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

 Instr

-0.78

accompan

-0.71

 kicker

-0.66

 Highlands

-0.62

 respons

-0.59

 Pulitzer

-0.59

enegger

-0.59

 Narr

-0.58

Gap

-0.57

 empowering

-0.57

POSITIVE LOGITS

isoft

1.01

irez

0.88

ead

0.83

chwitz

0.83

ircraft

0.83

FML

0.80

jac

0.79

uscript

0.78

verage

0.77

Rogue

0.76

Activations Density 0.648%

capitalized words

No Comments

No Known Activations