INDEX

Explanations

words related to controversy or contentious subjects

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

 Slug

-0.65

ELD

-0.63

eur

-0.61

 aids

-0.58

 Source

-0.58

 Able

-0.58

 Associated

-0.57

 Contributions

-0.56

 Accessed

-0.56

 asset

-0.55

POSITIVE LOGITS

 clarify

0.97

 assume

0.94

 proceed

0.92

 summarize

0.91

 pretend

0.87

 presume

0.87

 resolve

0.86

 discuss

0.86

 begin

0.86

nyder

0.86

Activations Density 0.012%

words related to controversy or contentious subjects

No Comments

No Known Activations