INDEX

Explanations

commands or instructions related to programming

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

sis

-0.88

>>\

-0.74

Merit

-0.72

abet

-0.71

fur

-0.69

kaya

-0.69

si

-0.66

vic

-0.64

present

-0.64

ple

-0.64

POSITIVE LOGITS

 them

1.14

 something

0.98

him

0.96

 these

0.93

 some

0.93

 another

0.91

 somebody

0.87

 someone

0.86

 THEM

0.85

 lots

0.84

Activations Density 2.297%

commands or instructions related to programming

No Comments

No Known Activations