INDEX

Explanations

geographical locations, particularly in British Columbia

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

kson

-0.93

nesota

-0.90

hammad

-0.85

ĺħ

-0.84

¥µ

-0.84

atown

-0.83

neapolis

-0.83

mble

-0.83

ographed

-0.81

stice

-0.81

POSITIVE LOGITS

 Columb

1.05

 Isles

1.03

 Sussex

0.98

 Columbia

0.97

 Yard

0.93

 Virgin

0.90

bridge

0.90

 Parliament

0.90

loo

0.88

 Buckingham

0.86

Activations Density 2.512%

geographical locations, particularly in British Columbia

No Comments

No Known Activations