INDEX

Explanations

instances of contrasting information or opposing viewpoints

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

tnc

-0.78

SPONSORED

-0.76

INS

-0.72

bas

-0.69

illet

-0.68

yz

-0.67

hig

-0.65

bre

-0.65

ursed

-0.65

bart

-0.64

POSITIVE LOGITS

 acknowledging

1.09

 technically

0.92

 conced

0.90

 researching

0.85

 admittedly

0.85

 admitting

0.79

 initially

0.77

 browsing

0.74

 agreeing

0.74

 discussing

0.73

Activations Density 1.037%

instances of contrasting information or opposing viewpoints

No Comments

No Known Activations