INDEX

Explanations

prolific speakers or sources of information in news articles

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_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

vier

-0.74

aden

-0.73

involved

-0.72

oute

-0.69

wart

-0.65

utz

-0.64

ught

-0.61

vent

-0.61

proc

-0.61

ocket

-0.61

POSITIVE LOGITS

 apart

0.77

 engagements

0.75

 excerpts

0.73

 anonymously

0.71

 Speaking

0.71

 fluent

0.70

 frankly

0.68

 unden

0.67

 Pitch

0.66

 Languages

0.66

Activations Density 0.027%

prolific speakers or sources of information in news articles

No Comments

No Known Activations