INDEX

Explanations

statistics and opinions related to political preferences and surveys

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

 Rampage

-0.63

 Gentleman

-0.61

 scalable

-0.59

 Cooking

-0.57

 hostage

-0.57

 Humane

-0.56

 billions

-0.56

 playbook

-0.55

 Fare

-0.55

Mug

-0.54

POSITIVE LOGITS

ecided

0.85

 undecided

0.78

 respondents

0.77

ogether

0.77

 placebo

0.77

OTAL

0.76

MpServer

0.76

).[

0.75

%).

0.74

 apiece

0.72

Activations Density 0.238%

statistics and opinions related to political preferences and surveys

No Comments

No Known Activations