INDEX

Explanations

positive, negative or neutral adjectives followed by adjectives or nouns

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

ĸļ

-1.18

emale

-0.85

ICLE

-0.81

 Exactly

-0.79

bsite

-0.77

hyde

-0.76

 TAMADRA

-0.76

uilding

-0.74

athlon

-0.72

isine

-0.70

POSITIVE LOGITS

 surprises

1.11

 goodies

1.01

 caveats

1.00

 tricks

0.98

 shenanigans

0.93

 gems

0.90

 quirks

0.89

 pointers

0.89

 bumps

0.88

 tweaks

0.86

Activations Density 0.281%

positive, negative or neutral adjectives followed by adjectives or nouns

No Comments

No Known Activations