INDEX

Explanations

words related to buying and selling items, particularly in an illicit or unethical manner

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

eworks

-0.68

atever

-0.63

Dys

-0.59

 Investigations

-0.57

 Buckley

-0.56

lav

-0.56

ulhu

-0.56

 Cthulhu

-0.56

 rehears

-0.56

 Goddard

-0.56

POSITIVE LOGITS

 prices

1.17

 price

1.13

 priced

1.02

Price

1.02

 Prices

0.95

 selling

0.94

 shelves

0.93

 sellers

0.91

 sale

0.90

price

0.90

Activations Density 1.996%

words related to buying and selling items, particularly in an illicit or unethical manner

No Comments

No Known Activations