INDEX

Explanations

mentions of written content, such as documentation or reports

oai_token-act-pair · gpt-3.5-turbo

articles such as "a" and "an" in the text

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

TDs

-0.72

 outweigh

-0.71

 colonists

-0.69

weights

-0.69

onto

-0.67

 actionGroup

-0.67

 combatants

-0.67

plates

-0.66

izons

-0.65

itiz

-0.64

POSITIVE LOGITS

 nutshell

1.24

 wake

0.84

 perverse

0.83

 twist

0.83

 statement

0.79

 brief

0.78

 bizarre

0.77

 footnote

0.76

 flurry

0.75

 courtroom

0.74

Activations Density 0.090%

mentions of written content, such as documentation or reports

articles such as "a" and "an" in the text

No Comments

No Known Activations