qwen2.5-coder-0.5b-everywhere-exact-w8

Anomly/qwen2.5-coder-0.5b-everywhere-exact-w8

qwen2.5-coder-0.5b-everywhere-exact-w8 is a machine learning model from Anomly. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.

Downloads · 30 days

8% of all-time downloads

All-time downloads

237

Public

Repo size

609 MB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.gguf516 MB · 77%

At a glance

License: apache-2.0
Access: Public
Created: Sep 6, 2026
Updated: Sep 6, 2026
SHA: 2095ca19

Base models

Qwen/Qwen2.5-Coder-0.5B-Instruct

Model card

From the Hugging Face model README

Anomly/qwen2.5-coder-0.5b-everywhere-exact-w8

The 0.5B sibling of Anomly/qwen2.5-coder-1.5b-everywhere-exact-w8: Qwen2.5-Coder-0.5B-Instruct, same execution-verified data, same exact-8-bit LoRA recipe, same receipts. It exists because a 2 GB ARM board (Ultra96-V2) runs it at a usable speed; the full write-up, the board and the verification steps are on the 1.5B page and apply unchanged (file names here use the student05_fixed8 row).

	HumanEval	HumanEval+	MBPP	MBPP+
exact 8-bit fine-tune, greedy, receipted	57.3	53.0	54.0	44.7

Identity (whole-graph dump SHA-256, canonical prompt): x86 96edc94d772072a9ba83dc8021c8fa0296b50e972a941a03d4cb84044662dbce, RTX 5090 96edc94d772072a9ba83dc8021c8fa0296b50e972a941a03d4cb84044662dbce, Ultra96 A53 96edc94d772072a9ba83dc8021c8fa0296b50e972a941a03d4cb84044662dbce; Blackhole 2688/2688 rows, ACCEPT; Alveo U200 2688/2688 rows, ACCEPT. The dumps themselves are in dumps/.

File: student05_fixed8-bposit8.gguf (SHA-256 1bd1e56e05be9ca30cfae77aa25c9898ab5cccbc51e0a61297164fda2463786b). Licence: Apache-2.0, as the upstream Qwen2.5-Coder-0.5B-Instruct.

License: apache-2.0
Created: Sep 6, 2026
Updated: Sep 6, 2026