INDEX

Explanations

mentions of named individuals

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

geries

-0.71

 Schiff

-0.68

 steroid

-0.59

raints

-0.59

 intervening

-0.58

 Bronze

-0.57

ocious

-0.55

 Slav

-0.55

 Slim

-0.55

 Goodell

-0.54

POSITIVE LOGITS

 deems

1.13

 deem

1.07

 sorely

0.87

 dearly

0.85

 cherish

0.85

 considers

0.79

 despise

0.78

 aspire

0.78

 termed

0.77

 encounter

0.75

Activations Density 1.821%

mentions of named individuals

No Comments

No Known Activations