INDEX

Explanations

repetitive patterns starting with "Another"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Another," indicating a focus on reiterating or introducing additional points

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hips

-0.90

ouls

-0.85

onies

-0.82

olas

-0.82

bows

-0.74

alties

-0.73

opus

-0.71

riages

-0.69

ships

-0.68

eries

-0.68

POSITIVE LOGITS

 example

0.95

worldly

0.94

 aspect

0.94

 pecul

0.91

 drawback

0.90

 thing

0.88

 reason

0.88

 avenue

0.87

 notable

0.87

 notch

0.86

Activations Density 0.046%

repetitive patterns starting with "Another"

instances of the word "Another," indicating a focus on reiterating or introducing additional points

No Comments

No Known Activations