INDEX

Explanations

references to fairness, justice, and equality

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

PROV

-0.81

aneous

-0.70

arij

-0.66

 Norn

-0.65

ATED

-0.65

Assembly

-0.63

pta

-0.61

ulous

-0.61

PE

-0.58

odied

-0.58

POSITIVE LOGITS

 Weasley

0.93

ships

0.88

fort

0.85

iton

0.84

lyn

0.84

nell

0.83

Ñĭ

0.80

itudes

0.79

ies

0.79

\\\\\\\\

0.79

Activations Density 3.335%

references to fairness, justice, and equality

No Comments

No Known Activations