Jinx
|
278d87286a
|
Remove broken in-place replacement code, use _tokenize_with_special_tokens only
|
2026-04-10 17:35:09 +00:00 |
|
Jinx
|
ca50973065
|
Fix tokenization: replace text token sequences with actual special token IDs
|
2026-04-10 17:28:51 +00:00 |
|
Jinx
|
d3b5f04f88
|
Add embed_tokens to LoRA targets + token ID verification before training
|
2026-04-10 17:20:06 +00:00 |
|
Jinx
|
f46995690c
|
Update runbook for tool-call token training run
|
2026-04-10 17:14:57 +00:00 |
|
Jinx
|
af497eb16c
|
Add training plan: teach model to emit native tool-call tokens
|
2026-04-10 17:07:28 +00:00 |
|
Jinx
|
d1e8c306e3
|
Add critical training objective: teach model to emit native tool-call tokens
|
2026-04-10 16:52:09 +00:00 |
|
Jinx
|
eb0850bca6
|
Fix: correct source dataset tag names for tool_call/tool_response
|
2026-04-10 06:37:43 +00:00 |
|
Jinx
|
99481ca127
|
Add HF_TOKEN env var
|
2026-04-10 06:33:09 +00:00 |
|
Jinx
|
9e723d393e
|
Fix: remove unsupported --model arg from prepare_data.py call
|
2026-04-10 06:32:55 +00:00 |
|
Jinx
|
adbd85366b
|
init commit
|
2026-04-10 06:24:05 +00:00 |
|
Jinx
|
46a3ddbb25
|
Add deployment runbook
|
2026-04-10 05:28:30 +00:00 |
|
Jinx
|
6af62c85d5
|
Add docker-compose with /srv mounts for persistent data
|
2026-04-10 05:21:04 +00:00 |
|
Jinx
|
82348341b0
|
Initial LoRA training setup for SmolLM3-3B tool calling
|
2026-04-10 05:11:05 +00:00 |
|