INDEX

Explanations

words related to reputation and credibility

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

cise

-0.86

pheus

-0.77

wed

-0.76

tein

-0.71

vention

-0.70

phone

-0.68

nes

-0.67

stad

-0.66

err

-0.66

apor

-0.65

POSITIVE LOGITS

 tarn

0.98

 reputation

0.95

 bearer

0.85

worthiness

0.79

ously

0.76

arily

0.76

 tremend

0.75

 internationally

0.73

 preced

0.72

 abroad

0.71

Activations Density 0.045%

words related to reputation and credibility

No Comments

No Known Activations