INDEX

Explanations

phrases related to headwear and clothing items

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 citiz

-0.75

ITNESS

-0.74

 newsp

-0.73

 practition

-0.73

 Accountability

-0.73

KNOWN

-0.71

GoldMagikarp

-0.71

Reviewer

-0.70

Qual

-0.70

 adolesc

-0.66

POSITIVE LOGITS

mith

1.30

poons

1.25

cale

1.25

creen

1.15

etting

1.13

hell

1.13

poon

1.12

avers

1.11

hare

1.10

ucker

1.08

Activations Density 0.226%

phrases related to headwear and clothing items

No Comments

No Known Activations