INDEX

Explanations

phrases related to legal matters and actions taken by individuals or groups

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 conveyed

-0.70

 transmitted

-0.68

Ĳ

-0.68

 attracted

-0.67

few

-0.65

 compared

-0.64

lied

-0.63

 preceded

-0.63

ially

-0.61

achy

-0.61

POSITIVE LOGITS

 grips

1.13

 fruition

0.95

 orbit

0.86

 shore

0.82

 town

0.80

 Dover

0.80

 prominence

0.79

 Hogwarts

0.79

 heights

0.78

 Vegas

0.76

Activations Density 3.886%

phrases related to legal matters and actions taken by individuals or groups

No Comments

No Known Activations