INDEX

Explanations

terms related to politics, international relations, and business transactions

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

vernment

-0.68

hene

-0.65

 concludes

-0.59

roxy

-0.59

onz

-0.59

 RELEASE

-0.58

 Yourself

-0.56

 Period

-0.55

Rew

-0.55

 ILCS

-0.55

POSITIVE LOGITS

 unaffected

0.88

 predominant

0.82

 excluded

0.78

 spons

0.77

 suffice

0.74

 likewise

0.74

 prominently

0.74

 abound

0.73

 incorpor

0.72

 onboard

0.71

Activations Density 3.253%

terms related to politics, international relations, and business transactions

No Comments

No Known Activations