INDEX

Explanations

phrases related to editors, publications, and journalistic activities

oai_token-act-pair · gpt-3.5-turbo

mentions of editorial roles or positions

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

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

 Simulator

-0.80

 simulator

-0.72

 arming

-0.67

 logging

-0.66

 annex

-0.65

 chamber

-0.62

 Chamber

-0.61

istries

-0.61

 awakening

-0.60

 earthqu

-0.60

POSITIVE LOGITS

 Editors

1.93

 editors

1.43

wikipedia

0.71

Kal

0.70

Sah

0.68

Sov

0.68

Summer

0.67

Ly

0.67

assetsadobe

0.66

 scales

0.66

Activations Density 0.015%

phrases related to editors, publications, and journalistic activities

mentions of editorial roles or positions

No Comments

No Known Activations