INDEX

Explanations

descriptions related to political figures and government events

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

natureconservancy

-0.80

ertain

-0.75

omorph

-0.73

itiveness

-0.70

erers

-0.68

omorphic

-0.67

 INCLUD

-0.67

 interacts

-0.66

Reviewer

-0.66

ulum

-0.65

POSITIVE LOGITS

 February

1.26

 December

1.24

 September

1.22

 August

1.21

 July

1.20

 October

1.20

 January

1.20

 March

1.18

 November

1.18

 June

1.18

Activations Density 6.908%

descriptions related to political figures and government events

No Comments

No Known Activations