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

ãĥĩãĤ£

-0.74

ãĥ¯

-0.73

ãĤ¨ãĥ«

-0.71

 span

-0.66

 wings

-0.65

)]

-0.62

defined

-0.62

(<

-0.61

 hail

-0.61

chance

-0.61

POSITIVE LOGITS

uana

0.77

emet

0.73

iners

0.72

rosse

0.69

alion

0.68

Mech

0.68

ERC

0.68

ogun

0.67

Beat

0.66

ufact

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.