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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

renheit

-0.76

Rain

-0.72

 lane

-0.69

vec

-0.68

anova

-0.67

epad

-0.66

ema

-0.65

resso

-0.65

TextColor

-0.64

aren

-0.64

POSITIVE LOGITS

 supra

0.75

 Firm

0.73

 partName

0.70

 descending

0.65

 Democr

0.62

uble

0.62

 Demonic

0.60

 Î±

0.59

 thereof

0.58

 tending

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.