INDEX

Explanations

words related to written rules or guidelines like "policy"

oai_token-act-pair · gpt-3.5-turbo

references to policies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Stain

-0.78

ITNESS

-0.73

ingly

-0.73

iculture

-0.72

amate

-0.71

Ago

-0.70

Sic

-0.67

ergy

-0.66

ymes

-0.65

azar

-0.64

POSITIVE LOGITS

holders

1.12

holder

1.09

 makers

1.06

making

1.05

 prescriptions

0.95

maker

0.89

 prohibiting

0.89

makers

0.87

 enforcement

0.85

 directives

0.84

Activations Density 0.047%

words related to written rules or guidelines like "policy"

references to policies

No Comments

No Known Activations