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

 resil

-0.81

natureconservancy

-0.80

atorium

-0.74

raltar

-0.68

 beaches

-0.64

 Sturgeon

-0.64

 rods

-0.63

 cryptoc

-0.61

hyde

-0.61

 dors

-0.59

POSITIVE LOGITS

ventory

0.88

tering

0.77

tered

0.72

lets

0.70

ror

0.69

 Principal

0.66

Side

0.66

 Dresden

0.65

------------------------------------------------

0.65

izations

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.