INDEX

Explanations

names and titles of people

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

enhagen

-0.71

ãĥ¯

-0.68

 Beware

-0.63

imates

-0.61

 flush

-0.61

ãģ®ç

-0.61

 multiplier

-0.60

chers

-0.59

chio

-0.59

 scales

-0.58

POSITIVE LOGITS

ospital

1.24

azard

1.17

ISTORY

1.15

anging

1.14

ttp

1.13

ometown

1.11

annah

1.11

awk

1.11

arrison

1.07

ottest

1.05

Activations Density 1.755%

names and titles of people

No Comments

No Known Activations