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

 Franch

-0.67

geoning

-0.65

ista

-0.64

igree

-0.63

poons

-0.62

 Investment

-0.62

erville

-0.61

oman

-0.59

ibaba

-0.59

bernatorial

-0.59

POSITIVE LOGITS

peak

0.71

translation

0.70

aber

0.70

gow

0.65

gy

0.65

sense

0.64

skirts

0.63

gged

0.62

history

0.61

)--

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.