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

 Clash

-0.81

-+-+

-0.76

GG

-0.74

Crash

-0.73

 Pixie

-0.72

ãĥ¼ãĥĨ

-0.70

RAG

-0.69

 Carbuncle

-0.69

ãĥ¼ãĥĨãĤ£

-0.69

AMA

-0.69

POSITIVE LOGITS

osis

0.70

oma

0.67

 invention

0.65

sin

0.65

 centers

0.65

ascus

0.64

 sciences

0.63

 paraph

0.62

 consec

0.61

 centres

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.