INDEX

Explanations

mentions of bladed items like swords, knives, or lightsabers

oai_token-act-pair · gpt-3.5-turbo

terms related to blades

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ADRA

-0.87

quickShipAvailable

-0.80

gdala

-0.78

bos

-0.76

riad

-0.75

ravel

-0.73

ally

-0.72

Particip

-0.71

ordable

-0.70

upon

-0.68

POSITIVE LOGITS

 blades

1.24

 blade

1.13

tips

1.09

blade

0.98

 Runner

0.95

 Cutter

0.91

 scissors

0.90

tip

0.90

 knives

0.89

 wielded

0.87

Activations Density 0.011%

mentions of bladed items like swords, knives, or lightsabers

terms related to blades

No Comments

No Known Activations