INDEX

Explanations

words related to gaining or increase

oai_token-act-pair · gpt-3.5-turbo

instances of the word "gained" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

pled

-0.79

iso

-0.69

align

-0.67

 Mole

-0.65

mate

-0.61

phis

-0.60

please

-0.59

opers

-0.58

mun

-0.57

pl

-0.57

POSITIVE LOGITS

 gained

1.06

ĸļ

0.92

 gains

0.92

 notoriety

0.84

 gain

0.84

 veter

0.84

iculty

0.80

 traction

0.77

icult

0.76

itive

0.75

Activations Density 0.006%

words related to gaining or increase

instances of the word "gained" in various contexts

No Comments

No Known Activations