INDEX

Explanations

phrases related to progress or steps in a process

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

å¥³

-0.74

aez

-0.74

 constitu

-0.70

IZE

-0.69

cci

-0.68

oho

-0.66

orio

-0.66

esian

-0.64

ãĥ³ãĤ¸

-0.64

email

-0.63

POSITIVE LOGITS

noon

1.29

wards

1.13

math

1.08

 completing

1.01

market

0.99

words

0.97

ward

0.96

 awhile

0.92

thought

0.92

 finishing

0.91

Activations Density 0.433%

phrases related to progress or steps in a process

No Comments

No Known Activations