INDEX

Explanations

locations or geographical names

oai_token-act-pair · gpt-3.5-turbo

mentions of specific geographical locations and significant individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Clever

-0.74

 Aether

-0.73

uality

-0.70

het

-0.65

otle

-0.65

 Alchemist

-0.63

 Proceed

-0.62

âĹ¼

-0.62

****************

-0.61

ask

-0.59

POSITIVE LOGITS

ruits

0.94

enhagen

0.92

ornia

0.90

 Fighters

0.90

selage

0.87

letcher

0.86

luent

0.85

iture

0.85

iltration

0.84

rican

0.83

Activations Density 0.056%

locations or geographical names

mentions of specific geographical locations and significant individuals

No Comments

No Known Activations