INDEX

Explanations

numerical values related to dates, numbers, and codes

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

Not in Any Lists

No Comments

Negative Logits

isman

-0.65

etimes

-0.65

aned

-0.64

 scrut

-0.64

 endors

-0.63

 knowledgeable

-0.63

ulously

-0.62

emade

-0.60

 snowball

-0.60

 relentless

-0.59

POSITIVE LOGITS

1.07

1.06

1.03

1.02

1.01

Activations Density 2.103%

numerical values related to dates, numbers, and codes

No Comments

No Known Activations