INDEX

Explanations

text related to advanced technical content with a focus on programming, data processing, and cryptography

oai_token-act-pair · gpt-3.5-turbo

concepts related to rules and exceptions

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

 Bron

-0.67

 Bunker

-0.65

 fundra

-0.62

 Nurs

-0.61

 Sierra

-0.61

 Rudd

-0.60

 Journalism

-0.60

 Amar

-0.60

 Hosp

-0.60

 Pulitzer

-0.59

POSITIVE LOGITS

 predetermined

0.82

teness

0.73

 matrix

0.73

 gradient

0.71

 respective

0.69

 observable

0.69

 desired

0.68

 occurrence

0.66

 immutable

0.66

?".

0.66

Activations Density 1.992%

text related to advanced technical content with a focus on programming, data processing, and cryptography

concepts related to rules and exceptions

No Comments

No Known Activations