INDEX

Explanations

comparative phrases denoting higher values or amounts

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

ĺħ

-0.74

 Recording

-0.66

xtap

-0.65

 Commissioner

-0.64

 Shooter

-0.64

 Roller

-0.63

 Record

-0.62

 Pair

-0.61

 Camel

-0.61

 Fist

-0.61

POSITIVE LOGITS

 than

1.22

 stringent

1.00

 importantly

0.98

than

0.97

 nuanced

0.97

 sophisticated

0.96

 mundane

0.94

 elaborate

0.94

 complicated

0.93

 expensive

0.89

Activations Density 0.110%

comparative phrases denoting higher values or amounts

No Comments

No Known Activations