INDEX

Explanations

past tense verbs ending in '-ed' referring to demonstrating achievement or success

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

 newsletters

-0.78

arta

-0.77

letal

-0.76

ataka

-0.69

umbn

-0.68

ades

-0.68

adish

-0.68

osate

-0.65

 Pastebin

-0.63

lished

-0.62

POSITIVE LOGITS

 invaluable

0.78

\\\\\\\\

0.77

 incapable

0.76

manship

0.72

 resilient

0.71

reader

0.70

stal

0.69

 ineffective

0.69

ance

0.68

 Chimera

0.67

Activations Density 0.061%

past tense verbs ending in '-ed' referring to demonstrating achievement or success

No Comments

No Known Activations