INDEX

Explanations

scientific and technical terms related to physics, economics, and academia

oai_token-act-pair · gpt-3.5-turbo

terms related to rankings and classifications, particularly in a business or competitive context

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 diligence

-0.72

 bureau

-0.71

 vigilance

-0.70

 nursing

-0.70

 forestry

-0.69

 coffin

-0.67

 trove

-0.67

 vigil

-0.67

 crypt

-0.66

 reel

-0.66

POSITIVE LOGITS

 Ones

1.16

 Month

1.12

 Awareness

1.10

 Ages

1.10

 Countries

1.10

 Characters

1.10

 Meal

1.08

 Position

1.08

 Enemy

1.07

 Replacement

1.06

Activations Density 0.497%

scientific and technical terms related to physics, economics, and academia

terms related to rankings and classifications, particularly in a business or competitive context

No Comments

No Known Activations