INDEX

Explanations

instances where belonging or interacting within a specific group or organization is emphasized

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to legal relationships and affiliations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

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_post

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

watching

-0.68

ngth

-0.67

ãĤ£

-0.66

heimer

-0.65

Attempt

-0.64

thank

-0.64

ãĤ»

-0.64

undrum

-0.61

ÄŁ

-0.61

pection

-0.60

POSITIVE LOGITS

 either

1.12

 whichever

1.00

 other

1.00

 different

1.00

 another

0.99

 reputable

0.93

 specific

0.90

 certain

0.85

 separate

0.83

 distinct

0.82

Activations Density 0.422%

instances where belonging or interacting within a specific group or organization is emphasized

references to legal relationships and affiliations

No Comments

No Known Activations