INDEX

Explanations

No Explanations Found

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

 record

-0.70

 dupl

-0.67

hower

-0.64

 populate

-0.62

 pornography

-0.62

 dominate

-0.61

 trash

-0.60

 anti

-0.59

gag

-0.59

 countless

-0.58

POSITIVE LOGITS

soType

0.85

quickShipAvailable

0.80

 Said

0.79

itates

0.77

ctuary

0.74

hani

0.74

eah

0.72

Pi

0.71

ruff

0.70

ebus

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.