INDEX

Explanations

conjunctions, specifically the word "and."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Detail

-0.61

 Rooms

-0.60

 Canter

-0.60

 Caucasus

-0.60

 Directorate

-0.60

 Integrity

-0.60

 Serie

-0.59

ADVERTISEMENT

-0.58

 absence

-0.57

Goo

-0.57

POSITIVE LOGITS

 participated

0.97

 interacted

0.93

 photographed

0.92

 kissed

0.90

 worshipped

0.89

 starred

0.87

 toured

0.87

 spoke

0.86

rogen

0.85

 campaigned

0.85

Activations Density 0.167%

conjunctions, specifically the word "and."

No Comments

No Known Activations