INDEX

Explanations

adjectives and verbs related to investigation and crime

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

ettlement

-0.60

 Bravo

-0.59

avier

-0.57

olar

-0.56

Statistics

-0.56

claimer

-0.56

uracy

-0.56

ggles

-0.55

redo

-0.55

brush

-0.53

POSITIVE LOGITS

 Parenthood

0.76

 clinically

0.72

 locally

0.71

 universally

0.68

 anecd

0.66

Ð»

0.66

by

0.66

wik

0.65

ologically

0.65

 inconsist

0.62

Activations Density 0.906%

adjectives and verbs related to investigation and crime

No Comments

No Known Activations