Changes / Feeds / Qwen/Qwen3-14B-AWQ
Qwen/Qwen3-14B-AWQ
2 changes to a tracked key since 2025-05-01, newest first. Head as last read: 31c69efc29464b6bb0aee1398b5a7b50a99340c3.
A quantization of Qwen/Qwen3-14B: 280 layers re-encoded; of the 163 tensors left unquantized, 83 byte-identical to the base, 80 changed at the same dtype (all norm weights). Compared tensor by tensor from sha256 fingerprints: 31c69efc2946 against Qwen/Qwen3-14B at 40c069824f42. A hash is equal or not; it says nothing about what an edit does.
| Date | Advisory | What changed, and what it does |
|---|---|---|
| 2025-05-19 | x_NXMC-2025-0074 | the chat template text changed but rendered byte-identically on the three fixed conversations (system+user, multi-turn, tool call), all of which end on a user turn; a conversation ending on an assistant turn (training data, prefill) was not rendered (2025-05-19).61e595dd0f1d -> 1a6fe1ecf891 Open in app |
| 2025-05-01 | x_NXMC-2025-0073 breaking | EOS stop list 151645 to [151645,151643]; generation now also stops on 151643, and still stops on every token it stopped on before; generation_config.json BOS declared as 151643 (was unset); context window 65536 to 40960 tokens (2025-05-01).e49110c26e08 -> bac7b87017e3 Open in app |
Pin it, and let a bot move the pin
from_pretrained("Qwen/Qwen3-14B-AWQ", revision="31c69efc29464b6bb0aee1398b5a7b50a99340c3")
With the Renovate preset, a pull request moves this pin when the head moves, and links back here.
[](https://www.neurarch.com/changes-feeds/models/Qwen/Qwen3-14B-AWQ.html)