INDEX

Explanations

phrases related to laws, regulations, policies, and societal issues

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

•MSE minimising swap

No Comments

Negative Logits

know

-0.65

inav

-0.63

 encounters

-0.62

 wills

-0.62

iates

-0.61

 Moines

-0.61

 IMAGES

-0.61

 Telesc

-0.60

 senses

-0.60

sburg

-0.60

POSITIVE LOGITS

 supposed

1.18

 definitely

1.15

nt

1.12

rael

1.12

senal

1.09

 currently

1.09

not

1.08

 hereby

1.05

 considered

1.05

 indeed

1.05

Activations Density 3.745%

phrases related to laws, regulations, policies, and societal issues

No Comments

No Known Activations