INDEX

Explanations

terms related to political and social events and conflicts, particularly those involving governments, organizations, and international actions

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

ĸļ

-0.90

 Brist

-0.74

 diagrams

-0.70

 precincts

-0.69

 laure

-0.69

 wraps

-0.68

 descriptions

-0.67

 Brus

-0.67

 dances

-0.65

 addresses

-0.64

POSITIVE LOGITS

establishment

1.28

Semitic

1.21

social

1.20

government

1.18

Semitism

1.14

corruption

1.13

drug

1.12

immigrant

1.11

piracy

1.11

capitalist

1.08

Activations Density 0.359%

terms related to political and social events and conflicts, particularly those involving governments, organizations, and international actions

No Comments

No Known Activations