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

 contrace

-0.70

-----------

-0.66

FORE

-0.65

 viol

-0.61

sett

-0.61

 intrusive

-0.61

%%%%

-0.61

 Inspector

-0.60

 dism

-0.58

"""

-0.58

POSITIVE LOGITS

atching

0.76

onding

0.74

anes

0.69

æ

0.69

iston

0.69

 Carbuncle

0.68

iability

0.68

qi

0.67

eas

0.67

 Chocobo

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.