INDEX

Explanations

specific words or phrases related to searching for information or names

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

 Democr

-0.76

JUST

-0.73

minus

-0.73

oir

-0.72

 Proud

-0.69

orld

-0.68

CG

-0.67

hement

-0.67

 gladly

-0.66

Sorry

-0.66

POSITIVE LOGITS

 clues

1.27

 signs

0.97

 answers

0.95

 solutions

0.92

 alternatives

0.92

 loopholes

0.89

ById

0.85

 keywords

0.84

 vulnerabilities

0.84

 correlations

0.83

Activations Density 0.076%

specific words or phrases related to searching for information or names

No Comments

No Known Activations