INDEX

Explanations

terms related to traditional or common practices or attributes

oai_token-act-pair · gpt-3.5-turbo

references to typical or familiar characteristics and experiences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

arching

-0.67

iggurat

-0.65

 recently

-0.65

illions

-0.64

ements

-0.64

 forthcoming

-0.63

 strategically

-0.63

aciously

-0.63

olulu

-0.62

ensibly

-0.61

POSITIVE LOGITS

aceae

0.74

natureconservancy

0.72

hello

0.70

taboola

0.67

 greeting

0.67

idae

0.65

cloth

0.64

WT

0.63

single

0.62

Standard

0.61

Activations Density 0.390%

terms related to traditional or common practices or attributes

references to typical or familiar characteristics and experiences

No Comments

No Known Activations