INDEX

Explanations

first-person pronouns and verbs

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

 Millennium

-0.66

earch

-0.64

 Harm

-0.57

Gad

-0.54

 Amen

-0.53

 Innocent

-0.52

 Gulf

-0.51

htaking

-0.51

 Welch

-0.50

prem

-0.50

POSITIVE LOGITS

've

1.05

'll

0.94

'd

0.87

're

0.87

owe

0.81

 knew

0.79

 hadn

0.79

 spent

0.77

 confess

0.76

 bounced

0.76

Activations Density 0.637%

first-person pronouns and verbs

No Comments

No Known Activations