INDEX

Explanations

phrases related to items and categories in a list

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

pload

-0.79

 Cycling

-0.75

 Lithuania

-0.74

 Tanaka

-0.68

 Hiroshima

-0.67

 Colomb

-0.67

 Acer

-0.66

 Alberto

-0.65

frog

-0.65

 Jean

-0.64

POSITIVE LOGITS

 assorted

1.07

 goodies

0.94

 aspects

0.83

 forms

0.83

 factors

0.82

artifacts

0.82

 indic

0.81

 accessories

0.81

cellaneous

0.79

quickShipAvailable

0.79

Activations Density 0.711%

phrases related to items and categories in a list

No Comments

No Known Activations