INDEX

Explanations

phrases related to legal or political statements

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

 idols

-0.76

onto

-0.74

Tags

-0.73

plates

-0.73

ettes

-0.70

import

-0.69

Gener

-0.67

Mods

-0.65

Ingredients

-0.65

 Dates

-0.65

POSITIVE LOGITS

 statement

1.19

 nutshell

1.07

 briefing

0.96

bid

0.92

 joint

0.90

 keynote

0.89

statement

0.89

 telephone

0.89

 memo

0.89

 footnote

0.88

Activations Density 0.056%

phrases related to legal or political statements

No Comments

No Known Activations