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

Vu

-0.71

 professionally

-0.66

 exemplary

-0.64

 architecture

-0.62

GSL

-0.61

 foreseeable

-0.61

rov

-0.60

 undis

-0.60

 indeed

-0.60

POSITIVE LOGITS

 nodd

0.82

buster

0.75

atomic

0.74

nm

0.74

antha

0.73

NH

0.73

uckland

0.71

mare

0.70

ãĤ´ãĥ³

0.70

omore

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.