INDEX

Explanations

the action of starting or beginning

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate an action or process starting

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.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

 Rica

-0.62

quickShipAvailable

-0.62

enance

-0.61

 Compass

-0.60

abiding

-0.60

afety

-0.60

 Guardian

-0.59

 Explan

-0.58

soDeliveryDate

-0.58

grounds

-0.58

POSITIVE LOGITS

 accumulate

1.02

 scratch

0.90

 compensate

0.87

 explore

0.87

 settle

0.86

 dismantle

0.86

 unravel

0.83

 squeeze

0.82

 earn

0.82

 regain

0.82

Activations Density 0.067%

the action of starting or beginning

phrases that indicate an action or process starting

No Comments

No Known Activations