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

 Leilan

-0.81

ãĤ¨

-0.80

ãĥ¯ãĥ³

-0.80

ãĥĥãĤ¯

-0.76

ģĸ

-0.76

Guest

-0.75

aida

-0.73

ãĥĹ

-0.73

ãĥĨ

-0.72

ãĥ³ãĤ¸

-0.72

POSITIVE LOGITS

 applicants

0.67

feeding

0.67

 gestation

0.66

 enrol

0.61

 outweigh

0.61

 scouts

0.60

 claimants

0.59

 biggest

0.59

 distribution

0.58

 prospect

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.