INDEX

Explanations

names and professions of characters from different backgrounds

oai_token-act-pair · gpt-3.5-turbo

characters involved in storytelling or narratives

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

Pwr

-0.84

endif

-0.81

 endif

-0.77

cknow

-0.73

quished

-0.72

azeera

-0.70

ĪĴ

-0.70

sequent

-0.69

 understatement

-0.68

 accountable

-0.67

POSITIVE LOGITS

 overheard

1.07

 stumbled

1.00

 accidentally

0.95

 browsing

0.95

 wandered

0.95

 casually

0.92

 overhe

0.91

 stroll

0.90

 reminis

0.88

 randomly

0.87

Activations Density 0.510%

names and professions of characters from different backgrounds

characters involved in storytelling or narratives

No Comments

No Known Activations