INDEX

Explanations

references to superlatives related to world ranking

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

ippi

-0.97

veyard

-0.84

ifer

-0.77

ibility

-0.77

»Ĵ

-0.75

imon

-0.73

Mn

-0.71

edin

-0.71

iple

-0.69

oug

-0.68

POSITIVE LOGITS

wide

1.00

liness

0.92

building

0.88

 premiere

0.85

wake

0.83

 Wide

0.83

Wide

0.79

 championships

0.77

 renowned

0.76

 religions

0.75

Activations Density 0.459%

references to superlatives related to world ranking

No Comments

No Known Activations