File size: 913 Bytes
2d6303e
 
6805d16
2d6303e
 
6805d16
 
 
 
 
2d6303e
 
 
6805d16
2d6303e
6805d16
2d6303e
 
6805d16
 
 
2d6303e
 
 
6805d16
 
 
05ea558
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
# SpiderPortal v5

Recurrent Depth Transformer with MLA attention, Engram memory, and MoE.

## Architecture
- Dense: 250M params — 2 prelude + 6 recurrent + 2 coda
- MoE: 5.3B params — 32 experts, top-2, 1 shared expert/layer
- MLA (DeepSeek-V2 style, 10.7x KV compression)
- Engram memory @ layers 1,4
- LTI + ACT + LoRA

## Training

### Dense
```
MICRO_BATCH=42 SEQ_LEN=2048 TARGET_TOKENS=12400000000 python mythos-fineweb-dense.py
```

### MoE (from dense checkpoint)
```
MICRO_BATCH=28 SEQ_LEN=2048 TARGET_TOKENS=12400000000 TRITON_COMPILE=1 DENSE_CKPT=... python mythos-fineweb-moe.py
```

## Dataset
Tokenized FineWeb-Edu sample-10BT — raw uint32 LE tokens
- train_tokens.bin: 7.7B tokens, 29GB
- metadata.json
## Current Training (1B MoE)

Config: 16 experts | top-1 routing | intermediate=1024 | 6 layers | n_loops=1
Params: 997M (18% Engram / 82% MoE)
VRAM: 43GB | Throughput: 40K tok/s

### Run