INDEX

Explanations

phrases related to extents or quantities

oai_token-act-pair · gpt-3.5-turbo

the phrase "a whole" followed by a quantity or descriptor

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

nings

-0.73

claimed

-0.69

thumbnails

-0.68

 filings

-0.67

always

-0.67

ensis

-0.67

intent

-0.66

marks

-0.66

peak

-0.62

CB

-0.61

POSITIVE LOGITS

heartedly

1.17

lot

1.11

 bunch

1.08

hearted

0.96

 slew

0.95

 swath

0.91

dozen

0.88

 dozen

0.85

 heap

0.84

LOT

0.82

Activations Density 0.039%

phrases related to extents or quantities

the phrase "a whole" followed by a quantity or descriptor

No Comments

No Known Activations