INDEX

Explanations

tokens containing the letters "une"

oai_token-act-pair · gpt-3.5-turbo

the word "une" and its variations, indicating a focus on a specific context or theme related to it

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

loo

-0.79

etheless

-0.74

ories

-0.74

İĭ

-0.69

BILITIES

-0.69

pread

-0.68

draw

-0.68

¶ħ

-0.68

 teasp

-0.68

 subp

-0.67

POSITIVE LOGITS

arthed

0.96

une

0.94

lect

0.66

CLASSIFIED

0.64

 immunity

0.63

ISTER

0.62

lected

0.61

quist

0.61

 insign

0.60

nels

0.59

Activations Density 0.014%

tokens containing the letters "une"

the word "une" and its variations, indicating a focus on a specific context or theme related to it

No Comments

No Known Activations