INDEX

Explanations

locations or features related to hills

oai_token-act-pair · gpt-3.5-turbo

references to hills or elevated landforms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

arius

-0.72

 Visa

-0.71

ATA

-0.71

PLIC

-0.68

Magic

-0.67

ofi

-0.67

mates

-0.66

 Exper

-0.66

mate

-0.65

SIGN

-0.65

POSITIVE LOGITS

 hill

3.67

 hills

2.63

Hill

1.99

 Hill

1.84

hill

1.80

 mountain

1.61

 cliff

1.55

 slope

1.53

 ridge

1.52

 valley

1.49

Activations Density 0.012%

locations or features related to hills

references to hills or elevated landforms

No Comments

No Known Activations