INDEX

Explanations

names of people or entities and their associated topics or activities

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

holes

-0.69

cil

-0.66

ordes

-0.62

okers

-0.61

oute

-0.60

adle

-0.60

aunders

-0.59

obin

-0.59

/#

-0.58

lectic

-0.58

POSITIVE LOGITS

 week

1.09

 month

1.07

 ditch

1.04

 year

1.02

 rites

0.96

 night

0.95

 gasp

0.86

decade

0.84

yr

0.82

 minute

0.81

Activations Density 0.281%

names of people or entities and their associated topics or activities

No Comments

No Known Activations