INDEX

Explanations

legal terminology and mentions

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

hoff

-0.72

mble

-0.70

 Tate

-0.69

 Churchill

-0.68

 Seasons

-0.67

xual

-0.66

 Runner

-0.66

 Solitaire

-0.64

eful

-0.64

ilon

-0.64

POSITIVE LOGITS

istically

0.95

ising

0.90

 precedent

0.89

 enforcement

0.88

 recourse

0.85

 proceedings

0.82

 enforce

0.82

suit

0.81

suits

0.81

 liability

0.80

Activations Density 0.336%

legal terminology and mentions

No Comments

No Known Activations