INDEX

Explanations

locations or places related to news 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ornings

-0.81

ACP

-0.65

 Lazarus

-0.64

arine

-0.64

inates

-0.64

addy

-0.64

rica

-0.63

ees

-0.61

ournals

-0.61

 Saban

-0.60

POSITIVE LOGITS

wald

0.91

pick

0.87

burg

0.84

sel

0.83

ktop

0.81

holm

0.80

PER

0.79

hire

0.77

cream

0.76

alez

0.76

Activations Density 0.039%

locations or places related to news events

No Comments

No Known Activations