INDEX

Explanations

references to locations or entities, especially related to the US

oai_token-act-pair · gpt-3.5-turbo

punctuation marks and their frequency in contextual phrases

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

%:

-0.86

-0.77

 guiActiveUnfocused

-0.73

FactoryReloaded

-0.72

 guiActiveUn

-0.71

hess

-0.67

ãĤ¨ãĥ«

-0.66

eworthy

-0.66

rance

-0.66

ries

-0.65

POSITIVE LOGITS

ain

1.08

 dude

0.97

eh

0.97

huh

0.96

 yeah

0.91

isn

0.90

 folks

0.89

 alright

0.89

sir

0.88

ya

0.86

Activations Density 0.242%

references to locations or entities, especially related to the US

punctuation marks and their frequency in contextual phrases

No Comments

No Known Activations