INDEX

Explanations

mentions of potential negative impact or harm to various entities, such as businesses or international reputation

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

mad

-0.78

lich

-0.74

optional

-0.68

az

-0.66

ult

-0.63

Salt

-0.63

involved

-0.61

 Cosponsors

-0.60

 caveats

-0.60

asks

-0.60

POSITIVE LOGITS

 morale

1.35

 livelihood

1.21

 credibility

1.05

 havoc

1.03

 psyche

0.94

 innocent

0.90

 competitiveness

0.89

 morals

0.87

 dignity

0.85

 productivity

0.83

Activations Density 1.739%

mentions of potential negative impact or harm to various entities, such as businesses or international reputation

No Comments

No Known Activations