INDEX

Explanations

terms related to arms and armaments

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

 Sound

-0.72

ween

-0.71

 Vide

-0.71

Mos

-0.65

 Dough

-0.62

lucent

-0.61

reddits

-0.61

 Atmosp

-0.60

due

-0.60

actionDate

-0.59

POSITIVE LOGITS

chair

1.58

aments

1.42

ament

1.41

ageddon

1.36

chairs

1.27

ature

1.06

atures

1.04

illary

1.03

ando

0.98

strong

0.97

Activations Density 0.020%

terms related to arms and armaments

No Comments

No Known Activations