INDEX

Explanations

geographical locations and events

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

nai

-0.75

pring

-0.64

itionally

-0.62

 alright

-0.60

ãĥ´

-0.60

iet

-0.60

ain

-0.59

inates

-0.56

 âĻ

-0.55

 perk

-0.53

POSITIVE LOGITS

 Goldstein

0.72

there

0.69

 there

0.65

orsi

0.65

 Journalists

0.61

 Hoffman

0.60

tis

0.59

 Fundamental

0.57

atever

0.57

 attackers

0.56

Activations Density 0.243%

geographical locations and events

No Comments

No Known Activations