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

SOURCE

-0.80

TABLE

-0.75

IENT

-0.71

Beck

-0.69

aston

-0.67

RED

-0.66

prep

-0.66

Doug

-0.66

Jennifer

-0.65

riks

-0.65

POSITIVE LOGITS

 mesh

0.75

 masc

0.68

 dynasty

0.67

flation

0.65

 stabil

0.62

 ventilation

0.61

 sweep

0.61

Sph

0.60

 threaded

0.60

 cartridge

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.