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

aples

-0.81

rients

-0.69

ching

-0.67

ricular

-0.66

 gorilla

-0.66

cles

-0.62

hunt

-0.62

cone

-0.60

ilitating

-0.60

zl

-0.60

POSITIVE LOGITS

renheit

0.72

ogyn

0.68

Cth

0.66

soType

0.66

Wire

0.64

actionDate

0.63

 Commissioner

0.62

Twe

0.62

 Receiver

0.62

 Bernstein

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.