INDEX

Explanations

dates mentioned in text

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate time frames, particularly references to "last week" or "this week."

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

ãĤ¦ãĤ¹

-0.85

ãĤ´ãĥ³

-0.75

arily

-0.73

arah

-0.73

bryce

-0.67

alo

-0.67

FIX

-0.63

orship

-0.63

Probably

-0.60

ocon

-0.60

POSITIVE LOGITS

 however

1.02

 meanwhile

0.80

 citing

0.75

 moreover

0.67

 HuffPost

0.67

 researchers

0.67

 amid

0.66

 Attorney

0.66

 Wikileaks

0.66

 Forbes

0.65

Activations Density 0.141%

dates mentioned in text

phrases that indicate time frames, particularly references to "last week" or "this week."

No Comments

No Known Activations