INDEX

Explanations

names of specific places and people

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

Interstitial

-0.69

tenance

-0.68

 Staples

-0.66

cers

-0.64

croft

-0.64

 minded

-0.63

sers

-0.61

 outgoing

-0.59

ModLoader

-0.58

 foremost

-0.57

POSITIVE LOGITS

ebra

1.37

ombies

1.25

ombie

1.24

odiac

1.23

ERO

1.22

ooming

1.20

ealous

1.17

eros

1.09

iggurat

1.07

imbabwe

1.03

Activations Density 0.373%

names of specific places and people

No Comments

No Known Activations