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

MpServer

-1.12

âĺ

-0.89

NetMessage

-0.80

seless

-0.79

âĹ¼

-0.76

atively

-0.75

inary

-0.74

thinkable

-0.73

mented

-0.72

thy

-0.71

POSITIVE LOGITS

 question

0.69

 Maurit

0.64

 subpoena

0.64

 partner

0.64

 surrogate

0.64

 Reeves

0.63

 cavity

0.63

 Senegal

0.61

 Holder

0.61

 guess

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.