INDEX

Explanations

information related to electronics and technology products

oai_token-act-pair · gpt-3.5-turbo

references to specific products or devices and their features

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

izons

-0.77

 Perspective

-0.76

 Ashes

-0.68

 Exploration

-0.67

DragonMagazine

-0.65

 exploring

-0.64

«ĺ

-0.62

Untitled

-0.61

spective

-0.60

FO

-0.58

POSITIVE LOGITS

 emit

1.20

 degrade

1.12

 contain

1.07

 cost

1.04

 withstand

1.03

 interfere

0.96

 emits

0.96

 corro

0.96

 prevent

0.95

 penetrate

0.94

Activations Density 0.246%

information related to electronics and technology products

references to specific products or devices and their features

No Comments

No Known Activations