INDEX

Explanations

names of places and organizations

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

staking

-0.62

roman

-0.58

 forgiving

-0.56

lap

-0.53

 allowances

-0.53

 scratch

-0.52

borne

-0.52

pling

-0.51

 blazing

-0.51

 depress

-0.50

POSITIVE LOGITS

0.91

0.86

icz

0.85

opol

0.83

iak

0.82

0.81

Ã¥

0.81

vous

0.80

din

0.77

theless

0.76

Activations Density 1.156%

names of places and organizations

No Comments

No Known Activations