INDEX

Explanations

comparisons using the phrase "not as" to highlight differences in quality or extent

oai_token-act-pair · gpt-3.5-turbo

comparative expressions that convey limitations or shortcomings

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

 COUR

-0.72

 ACTIONS

-0.71

 TAMADRA

-0.65

 Macintosh

-0.63

YES

-0.61

NUM

-0.61

itudes

-0.60

YES

-0.60

*:

-0.59

onomy

-0.59

POSITIVE LOGITS

 anymore

0.89

yet

0.86

imilar

0.83

irlf

0.83

far

0.81

 much

0.80

vernment

0.79

pired

0.78

 well

0.75

 evidenced

0.74

Activations Density 0.084%

comparisons using the phrase "not as" to highlight differences in quality or extent

comparative expressions that convey limitations or shortcomings

No Comments

No Known Activations