INDEX

Explanations

mentions of reversals or reversions

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

DragonMagazine

-0.75

 Pebble

-0.74

BuyableInstoreAndOnline

-0.71

 Sturgeon

-0.64

 Nest

-0.64

 buckle

-0.62

 Sammy

-0.60

 Northern

-0.59

 Berm

-0.58

 Samoa

-0.58

POSITIVE LOGITS

becca

1.18

versible

1.12

brand

1.11

levant

1.10

porter

1.08

interpret

1.07

jected

1.07

venge

1.04

usable

1.04

cover

1.03

Activations Density 1.038%

mentions of reversals or reversions

No Comments

No Known Activations