INDEX

Explanations

instances where something is being taken advantage of or utilized effectively

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iva

-0.80

 dstg

-0.73

cade

-0.72

lime

-0.70

aughtered

-0.69

retty

-0.69

ura

-0.68

ribe

-0.68

usa

-0.68

gru

-0.68

POSITIVE LOGITS

 loopholes

1.28

 opportunities

1.12

 loophole

0.98

 weaknesses

0.95

 opportunity

0.95

 existing

0.91

 vulnerabilities

0.85

 Opportun

0.81

 strengths

0.79

 abundant

0.79

Activations Density 0.054%

instances where something is being taken advantage of or utilized effectively

No Comments

No Known Activations