INDEX

Explanations

conjunctions and prepositions

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

 Gleaming

-0.68

tein

-0.64

 Cock

-0.64

 Sierra

-0.62

 Dragonbound

-0.62

igraph

-0.61

 Dian

-0.61

hoe

-0.61

Hack

-0.61

Dou

-0.59

POSITIVE LOGITS

rogens

1.13

rogen

1.07

romeda

0.79

then

0.76

Else

0.74

nery

0.74

 parcel

0.73

hra

0.71

erity

0.71

thens

0.70

Activations Density 2.021%

conjunctions and prepositions

No Comments

No Known Activations