INDEX

Explanations

No Explanations Found

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

eline

-0.92

orah

-0.85

elle

-0.82

ded

-0.78

arcity

-0.77

entin

-0.76

gdala

-0.74

olon

-0.72

awan

-0.72

jing

-0.71

POSITIVE LOGITS

soever

0.66

 Poly

0.65

 Conditions

0.65

 Misc

0.64

 ILCS

0.64

ÙĲ

0.62

ABS

0.62

 Gram

0.61

 Nether

0.60

 bends

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.