INDEX

Explanations

locations or places mentioned in the text, such as states or countries

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

lopp

-0.69

onym

-0.67

adobe

-0.66

endum

-0.66

odox

-0.66

ãĥ¼ãĥĨãĤ£

-0.65

sense

-0.65

á½

-0.64

tk

-0.64

 Winnipeg

-0.61

POSITIVE LOGITS

 respectively

1.68

 alike

1.51

 combined

1.00

 coasts

0.93

 Empires

0.87

 economies

0.84

 kingdoms

0.81

 sectors

0.80

 campuses

0.80

 realms

0.78

Activations Density 0.164%

locations or places mentioned in the text, such as states or countries

No Comments

No Known Activations