INDEX

Explanations

conjunctions and phrases that express relationships or conditions between ideas

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Formation

-0.68

 Thoughts

-0.67

 Biology

-0.65

 wondering

-0.64

 moot

-0.62

 Borough

-0.61

 Anyway

-0.60

 Cards

-0.59

 Conference

-0.57

 guessing

-0.56

POSITIVE LOGITS

 satisfies

1.12

 incorporates

1.07

 resembles

1.03

 conform

1.01

 integrates

1.01

 utilizes

0.96

 lasts

0.94

 protects

0.93

 reflects

0.93

 fits

0.92

Activations Density 0.151%

conjunctions and phrases that express relationships or conditions between ideas

No Comments

No Known Activations