INDEX

Explanations

names of locations and names of people, especially those with unusual or less common spellings

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

ured

-0.70

ELS

-0.65

 Gazette

-0.61

ĸļ

-0.60

UME

-0.60

urities

-0.60

ers

-0.59

erest

-0.58

urer

-0.57

eneg

-0.56

POSITIVE LOGITS

lein

1.20

jee

1.18

geist

1.15

rors

1.10

idge

1.05

getic

1.03

bilt

1.03

wald

1.00

gency

0.99

ror

0.96

Activations Density 1.110%

names of locations and names of people, especially those with unusual or less common spellings

No Comments

No Known Activations