INDEX

Explanations

names, locations, and dates

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

idential

-0.69

 Dickinson

-0.69

CTR

-0.67

CoC

-0.66

yards

-0.64

flies

-0.64

 Arist

-0.63

ivities

-0.62

 Shades

-0.61

agne

-0.60

POSITIVE LOGITS

ceans

1.25

vernight

1.17

atmeal

1.13

culus

1.10

mbudsman

1.09

lymp

1.08

scill

1.07

mega

1.06

gee

1.01

cean

1.00

Activations Density 1.411%

names, locations, and dates

No Comments

No Known Activations