INDEX

Explanations

phrases related to management and project tasks

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

heim

-0.67

eka

-0.67

dt

-0.65

peria

-0.65

lat

-0.65

phal

-0.64

bal

-0.64

heimer

-0.62

 confir

-0.62

kie

-0.61

POSITIVE LOGITS

 backwards

0.68

irection

0.67

GGGGGGGG

0.67

 crap

0.67

 illegally

0.66

 themselves

0.66

redients

0.66

 stuff

0.66

 dangerously

0.65

 machines

0.65

Activations Density 0.343%

phrases related to management and project tasks

No Comments

No Known Activations