INDEX

Explanations

programming syntax and code-related activations

oai_token-act-pair · gpt-3.5-turbo

procedural or functional components of programming code

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

mble

-0.80

itness

-0.75

 nodd

-0.73

nuts

-0.71

nut

-0.68

elf

-0.67

ened

-0.67

enthal

-0.67

 takedown

-0.66

edded

-0.65

POSITIVE LOGITS

--+

0.86

âĶĢâĶĢ

0.85

00200000

0.80

OOL

0.78

advertising

0.74

ivas

0.74

PsyNetMessage

0.74

Constructed

0.71

ãħĭ

0.70

âĶĢâĶĢâĶĢâĶĢ

0.70

Activations Density 0.003%

programming syntax and code-related activations

procedural or functional components of programming code

No Comments

No Known Activations