INDEX

Explanations

words related to policies and actions

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

pmwiki

-0.76

ãĤ´ãĥ³

-0.75

 preferring

-0.74

onyms

-0.67

advertisement

-0.67

 msec

-0.66

 sidx

-0.64

Hmm

-0.62

 Originally

-0.62

sim

-0.61

POSITIVE LOGITS

rid

0.86

enance

0.84

uate

0.79

 mankind

0.79

 communities

0.75

 humankind

0.72

 people

0.71

balance

0.71

 them

0.69

 souls

0.68

Activations Density 0.435%

words related to policies and actions

No Comments

No Known Activations