INDEX

Explanations

details and discussions related to politics, policy, regulations, and personal health issues

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 predecessor

-0.72

 Rhodes

-0.71

SourceFile

-0.67

tein

-0.67

 Sisters

-0.66

staff

-0.64

HQ

-0.64

Ake

-0.62

 protocol

-0.60

 competitor

-0.59

POSITIVE LOGITS

 clam

1.12

 perceive

1.09

 perce

1.03

 thirst

0.91

 distrust

0.90

 instinctively

0.90

 hungry

0.89

 gull

0.87

 craving

0.86

 watching

0.86

Activations Density 2.748%

details and discussions related to politics, policy, regulations, and personal health issues

No Comments

No Known Activations