INDEX

Explanations

instances of words related to law, investigation, and accountability

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

ovember

-0.90

nesota

-0.78

busters

-0.74

Phones

-0.73

artifacts

-0.71

 dearly

-0.70

Stars

-0.69

izons

-0.69

ifles

-0.68

landish

-0.68

POSITIVE LOGITS

ly

0.96

ness

0.90

 dismantling

0.90

bred

0.87

 adherence

0.83

 lobbying

0.81

 understatement

0.81

 enough

0.79

lly

0.77

 examination

0.77

Activations Density 1.316%

instances of words related to law, investigation, and accountability

No Comments

No Known Activations