INDEX

Explanations

mentions of specific locations or events related to historical conflicts or events

oai_token-act-pair · gpt-3.5-turbo

terms related to confinement or imprisonment

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

keyes

-0.64

"))

-0.61

TeX

-0.61

 Sunshine

-0.60

 slate

-0.57

rium

-0.56

anse

-0.55

 POLITICO

-0.55

 Clojure

-0.54

 amen

-0.52

POSITIVE LOGITS

 shortly

0.87

 during

0.83

 last

0.83

 earlier

0.83

 centuries

0.80

 millennia

0.79

 whilst

0.78

 months

0.78

 circa

0.76

 smuggled

0.76

Activations Density 0.303%

mentions of specific locations or events related to historical conflicts or events

terms related to confinement or imprisonment

No Comments

No Known Activations