File size: 1,448 Bytes
ddfe382
 
 
 
 
 
 
7056f24
 
 
 
 
 
 
 
 
ddfe382
 
 
 
 
 
 
7056f24
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ddfe382
 
 
 
 
7056f24
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
37
38
39
40
41
42
43
44
45
46
47
48
49
# Dense Training Plan

## Setup

Use the same JSONL checkpoint bundle as MoE. Keep random seed, batch token
target, learning-rate schedule, and validation split aligned.

Primary input:

```text
dataset/<checkpoint>/dataset/*.jsonl
```

Do not train from hidden staging folders unless rebuilding checkpoints. Hidden
`data/curated_upload` is a staging queue, not the training contract.

## Curriculum

1. Start with short and medium records.
2. Add long FIM examples.
3. Add continuation/code generation records.
4. Evaluate FIM and next-token loss separately.

## H100 Defaults

- Precision: bf16.
- Context: 2048 smoke, 4096 main run.
- Optimizer: AdamW or fused AdamW.
- LR schedule: warmup then cosine decay.
- Gradient clipping: enabled.
- Batch target: choose by total tokens, not examples, because FIM lengths vary.
- Validation: fixed files or fixed checksums from the checkpoint report.

## Preprocessing

- Preserve whitespace exactly.
- Append `<|endoftext|>` between records during packing.
- Keep FIM markers visible; do not strip special tokens.
- Reject empty `text`, invalid JSONL, and records exceeding context unless
  deterministic chunking is enabled.
- Bucket by token length to reduce padding waste.

## Stop Conditions

- Validation loss diverges twice after LR reduction.
- FIM metrics regress while train loss improves.
- Tokenizer round-trip or special-token audit fails.
- Syntax parse rate drops while raw loss improves.