INDEX

Explanations

names and specific identifying terms

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

icult

-0.76

yrinth

-0.74

aunders

-0.74

EMS

-0.73

isexual

-0.73

romy

-0.71

istar

-0.69

iasm

-0.68

psey

-0.66

nuts

-0.65

POSITIVE LOGITS

plates

1.49

plate

1.34

paces

1.15

ames

0.95

paced

0.92

names

0.89

tag

0.87

 recognition

0.87

 aliases

0.86

akes

0.84

Activations Density 0.614%

names and specific identifying terms

No Comments

No Known Activations