INDEX

Explanations

terms related to acts of respect or recognition, especially in formal contexts

oai_token-act-pair · gpt-3.5-turbo

references to the concept of honor

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

arre

-0.96

 fielded

-0.70

roads

-0.67

Done

-0.67

hesive

-0.66

aghetti

-0.62

 Ange

-0.62

hover

-0.61

--------

-0.61

 glimps

-0.60

POSITIVE LOGITS

ably

1.11

abilia

1.02

 honors

1.00

ific

0.99

 honoured

0.88

 honored

0.88

 honor

0.83

 honoring

0.82

uum

0.81

 honour

0.77

Activations Density 0.023%

terms related to acts of respect or recognition, especially in formal contexts

references to the concept of honor

No Comments

No Known Activations