INDEX

Explanations

numbers and measurements that indicate loss or damage

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "loss" and its variations indicating various forms of loss

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ansky

-0.71

ENTS

-0.70

ECK

-0.69

ATT

-0.64

ricks

-0.62

 inventive

-0.60

abad

-0.60

kov

-0.60

raper

-0.59

 Caucas

-0.59

POSITIVE LOGITS

 aversion

0.93

iem

0.85

 incurred

0.85

 Loss

0.84

 loss

0.81

loss

0.75

luster

0.75

esville

0.74

front

0.73

lessly

0.72

Activations Density 0.023%

numbers and measurements that indicate loss or damage

occurrences of the word "loss" and its variations indicating various forms of loss

No Comments

No Known Activations