INDEX

Explanations

information related to political events, including statements from political figures and analysis of policies

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

 appropri

-0.82

 pudding

-0.74

 spir

-0.74

 elev

-0.72

 inund

-0.71

 tackling

-0.71

 bricks

-0.70

 unus

-0.70

 exemplary

-0.70

 veget

-0.69

POSITIVE LOGITS

âĢ¦)

1.43

Laughs

1.33

hide

1.25

laughs

1.22

...)

1.21

See

1.21

emphasis

1.19

Note

1.18

Courtesy

1.17

Side

1.14

Activations Density 0.290%

information related to political events, including statements from political figures and analysis of policies

No Comments

No Known Activations