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

ãĤĮ

-0.75

entimes

-0.75

acion

-0.73

een

-0.71

rs

-0.71

rats

-0.70

folios

-0.70

eur

-0.68

 congratulations

-0.68

Links

-0.67

POSITIVE LOGITS

 buck

0.74

soDeliveryDate

0.66

zeb

0.66

 hawk

0.63

 newcom

0.62

 little

0.61

 displacement

0.60

 Harper

0.59

 historian

0.59

 Hale

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.