INDEX

Explanations

words related to legal actions, government actions, and programs

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

quickShipAvailable

-0.81

soDeliveryDate

-0.71

 Explan

-0.68

 Coff

-0.67

furt

-0.66

cloth

-0.66

lights

-0.66

 meant

-0.65

mails

-0.64

 Respons

-0.63

POSITIVE LOGITS

 manipulate

1.24

 communicate

1.24

 withstand

1.21

 survive

1.20

 negotiate

1.19

 cope

1.19

 penetrate

1.18

 participate

1.17

 utilize

1.17

 fend

1.17

Activations Density 0.849%

words related to legal actions, government actions, and programs

No Comments

No Known Activations