INDEX

Explanations

instances of the word "highly" in various contexts

oai_token-act-pair · gpt-3.5-turbo

the word "highly" and its variations, indicating it is focused on emphasizing intensity or degree

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

ppo

-0.82

 Dare

-0.79

ories

-0.79

 Parenthood

-0.77

 Chaser

-0.76

 Nightmares

-0.74

 Ashe

-0.72

 Tycoon

-0.72

 Slug

-0.72

 Sins

-0.70

POSITIVE LOGITS

 regarded

0.94

 educated

0.90

 unlikely

0.88

 improbable

0.85

 influential

0.85

 susceptible

0.85

 interconnected

0.84

 rated

0.84

charged

0.82

 cultured

0.81

Activations Density 0.024%

instances of the word "highly" in various contexts

the word "highly" and its variations, indicating it is focused on emphasizing intensity or degree

No Comments

No Known Activations