INDEX

Explanations

government and legal terms related to law enforcement and administration

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

 contrasts

-0.71

 bets

-0.65

pires

-0.61

 eclips

-0.60

 dominates

-0.60

 spans

-0.59

 contrasted

-0.59

 masks

-0.59

 contrast

-0.59

pread

-0.58

POSITIVE LOGITS

 requesting

0.86

mbudsman

0.85

mbuds

0.82

 ASAP

0.80

 permission

0.75

istrate

0.75

 asking

0.74

 beforehand

0.74

 regarding

0.72

istries

0.72

Activations Density 0.374%

government and legal terms related to law enforcement and administration

No Comments

No Known Activations