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

regnancy

-0.86

isen

-0.79

IME

-0.77

URE

-0.73

ITNESS

-0.72

Ka

-0.69

ythm

-0.67

raid

-0.67

TPPStreamerBot

-0.67

ERN

-0.67

POSITIVE LOGITS

 Gord

0.76

 Amos

0.66

amy

0.65

***

0.63

 Willie

0.62

 Hubbard

0.61

DAC

0.61

ghai

0.60

Poc

0.60

 Rats

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.