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

aukee

-0.82

 unfocusedRange

-0.77

EMA

-0.76

NM

-0.71

eway

-0.71

hire

-0.70

raper

-0.70

$$$$

-0.70

CBC

-0.68

hyde

-0.67

POSITIVE LOGITS

 Creed

0.86

 Slav

0.68

 Dying

0.67

ostics

0.64

 Pont

0.61

ision

0.60

 Afgh

0.60

 Caucasus

0.59

Fir

0.58

 Spart

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.