INDEX

Explanations

actions related to advertising and brand controversies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

Sov

-0.73

aceae

-0.70

swick

-0.69

Cong

-0.69

inside

-0.68

cong

-0.67

progress

-0.65

albeit

-0.65

DragonMagazine

-0.64

quart

-0.64

POSITIVE LOGITS

 registrations

0.85

 abruptly

0.80

 offending

0.77

 refunds

0.76

 unnecessary

0.76

 altogether

0.75

 unres

0.75

 withdrawals

0.74

 indefinitely

0.71

owship

0.69

Activations Density 0.208%

actions related to advertising and brand controversies

No Comments

No Known Activations