INDEX

Explanations

specific codes, identifiers, and programming-related terms

oai_token-act-pair · gpt-3.5-turbo

references to generic categories or placeholders in information

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

 allerg

-0.76

 Haram

-0.71

 influenza

-0.62

 Transformers

-0.61

 Loren

-0.60

 Aden

-0.60

addons

-0.59

 Benghazi

-0.58

 Warcraft

-0.58

 blasphemy

-0.56

POSITIVE LOGITS

odox

0.82

leneck

0.80

Rap

0.73

Í

0.67

à¼

0.67

=~=~

0.65

utterstock

0.64

Mus

0.63

Ther

0.62

Activations Density 0.196%

specific codes, identifiers, and programming-related terms

references to generic categories or placeholders in information

No Comments

No Known Activations