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

ËĪ

-0.73

OD

-0.72

 COUR

-0.71

ivals

-0.71

 Yamato

-0.70

ãĤ¨ãĥ«

-0.70

 Emirates

-0.69

 Domin

-0.68

 Dynasty

-0.68

TRY

-0.67

POSITIVE LOGITS

ktop

0.70

 obsc

0.67

 overdose

0.66

 cannabin

0.64

minecraft

0.63

nyder

0.62

 lith

0.62

ijk

0.61

 reper

0.61

 arter

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.