INDEX

Explanations

words related to knowledge, awareness, and understanding

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

ItemTracker

-0.76

phrine

-0.74

otion

-0.72

isco

-0.72

acco

-0.69

onding

-0.69

erva

-0.69

berus

-0.68

avored

-0.68

thren

-0.67

POSITIVE LOGITS

ledge

1.20

ledged

1.16

how

1.06

 firsthand

1.03

 beforehand

0.98

lege

0.97

 instinctively

0.85

ingly

0.82

how

0.82

 whats

0.81

Activations Density 1.326%

words related to knowledge, awareness, and understanding

No Comments

No Known Activations