INDEX

Explanations

descriptions of violent incidents involving individuals in specific locations

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

 ensures

-0.70

 retains

-0.70

 contributes

-0.69

 viable

-0.69

 excludes

-0.68

 resists

-0.68

 fullest

-0.67

integ

-0.67

 effectiveness

-0.67

alties

-0.66

POSITIVE LOGITS

 overheard

1.07

 suddenly

1.00

 acquaintance

0.95

 fateful

0.94

 accidentally

0.93

iphany

0.93

Suddenly

0.89

 Suddenly

0.86

 noticing

0.85

 encountering

0.84

Activations Density 3.369%

descriptions of violent incidents involving individuals in specific locations

No Comments

No Known Activations