INDEX

Explanations

headlines or news reports from CNN indicated by the presence of "(CNN)" and a high activation value

oai_token-act-pair · gpt-3.5-turbo

occurrences of parentheses and their context, particularly related to CNN references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 assum

-0.70

zos

-0.67

ãĤ´

-0.65

ãĥĵ

-0.65

ãĥĦ

-0.64

fee

-0.64

spr

-0.64

 stabil

-0.62

cf

-0.61

antage

-0.60

POSITIVE LOGITS

 Latest

0.88

 Thousands

0.84

 Dozens

0.80

 Authorities

0.78

 Instead

0.78

 Amid

0.73

 Hundreds

0.72

 Speaking

0.72

 Parenthood

0.72

 Despite

0.71

Activations Density 0.030%

headlines or news reports from CNN indicated by the presence of "(CNN)" and a high activation value

occurrences of parentheses and their context, particularly related to CNN references

No Comments

No Known Activations