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

 Explan

-0.74

 Plain

-0.70

 Explain

-0.65

 Hass

-0.63

 Belief

-0.62

nir

-0.61

iser

-0.60

 Cosmos

-0.59

 Situation

-0.59

 totality

-0.59

POSITIVE LOGITS

away

0.77

iton

0.76

ENN

0.70

tern

0.70

enza

0.68

prus

0.68

è¦

0.67

rones

0.67

awei

0.66

UNE

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.