INDEX

Explanations

numbers written as words

oai_token-act-pair · gpt-3.5-turbo

patterns related to numerical values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

loo

-0.80

REAM

-0.73

WARD

-0.70

ffee

-0.70

hips

-0.70

Accessory

-0.66

hire

-0.64

 Denis

-0.63

OHN

-0.63

RAFT

-0.63

POSITIVE LOGITS

eral

0.98

emonic

0.94

num

0.91

pty

0.88

quist

0.87

Num

0.87

phys

0.84

BER

0.77

num

0.75

iatures

0.74

Activations Density 0.023%

numbers written as words

patterns related to numerical values

No Comments

No Known Activations