CoolFace
Modelpublic

mihailgribov/typecastlm-qwen3.5-3.8b

sourceHugging Faceapache-2.0updated 5h agoView on Hugging Face
0likes66downloads
49 commits on main
1d7ba9d5h ago

README.md: the model in every form in one repository

mihailgribov
5f6e9315h ago

reader.py: the model in every form in one repository

mihailgribov
b13be6b5h ago

prompt.json: the model in every form in one repository

mihailgribov
00bf35c5h ago

head.json: the model in every form in one repository

mihailgribov
35ec3605h ago

typecastlm-qwen3.5-3.8b-bf16.gguf

mihailgribov
9a4e77e5h ago

typecastlm-qwen3.5-3.8b-q8_0.gguf

mihailgribov
103978610h ago

remove typecastlm-qwen3.5-3.8b-bf16.gguf: staged in the private repository first

mihailgribov
5b09c4210h ago

remove typecastlm-qwen3.5-3.8b-q8_0.gguf: staged in the private repository first

mihailgribov
460b97b10h ago

typecastlm-qwen3.5-3.8b-bf16.gguf

mihailgribov
d37edd210h ago

typecastlm-qwen3.5-3.8b-q8_0.gguf

mihailgribov
dcfecfd3d ago

card: our own pools only; notice and prompt note follow

mihailgribov
a122c403d ago

NOTICE: the head is 39 rows now, and the checkpoint is sharded

mihailgribov
194b9523d ago

card: the affiliation note belongs at the bottom, not in the middle of the pitch

mihailgribov
b658cf23d ago

prompt.json: only the fields the readers actually use

mihailgribov
951d4373d ago

prompt.json: the system line JevBench was measured with

mihailgribov
fdd35e43d ago

prompt.json carries the marks wording; reader takes the stronger surface form

mihailgribov
2d325c73d ago

card: ten levels recommended, thirty-six accepted, and digits beat letters

mihailgribov
91990213d ago

reader: marks run A..Z, and the head is no longer described as 29 rows

mihailgribov
48eed453d ago

head carries A..Z; checkpoint split into shards so the head is 195 KB of them

mihailgribov
15c4c884d ago

Upload NOTICE with huggingface_hub

mihailgribov
bcd7e674d ago

Upload LICENSE with huggingface_hub

mihailgribov
b2ac2f64d ago

Upload reader.py with huggingface_hub

mihailgribov
da9525b4d ago

Upload README.md with huggingface_hub

mihailgribov
941cbd34d ago

calibration: where the values live

mihailgribov
12f04db4d ago

calibration: what it was fitted on

mihailgribov
45404564d ago

card: four modes

mihailgribov
e4de5f94d ago

card: three modes in the first screen

mihailgribov
5a8efe84d ago

card: describe the three modes

mihailgribov
7610b8a4d ago

card: the third answer up front

mihailgribov
bbf92d54d ago

card: drop what the NOTICE already states

mihailgribov
ea992f44d ago

card and prompt: values only

mihailgribov
89c2e494d ago

card and prompt: values only

mihailgribov
1fb98354d ago

card: measured numbers only

mihailgribov
3ac526b4d ago

release: three modes, per-mode temperatures, one-line install

mihailgribov
45170b44d ago

release: three modes, per-mode temperatures, one-line install

mihailgribov
7982b1f4d ago

the token lists are not part of the release

mihailgribov
0a612bd4d ago

the head is computed vectors, not trained ones

mihailgribov
5602aca4d ago

the head is computed vectors, not trained ones

mihailgribov
e66695d4d ago

the head recipe is not part of the release

mihailgribov
b49bf5b4d ago

describe what changed, not how the head was built

mihailgribov
310e5e04d ago

describe what changed, not how the head was built

mihailgribov
33679914d ago

notice: only the marks are copied rows; the three answers are computed from labelled data

mihailgribov
fa292fd4d ago

notice: only the marks are copied rows; the three answers are computed from labelled data

mihailgribov
31485c04d ago

licence: README.md

mihailgribov
1a35fd94d ago

licence: NOTICE

mihailgribov
0128b0a4d ago

licence: LICENSE

mihailgribov
1b34d6b4d ago

cut at block 27, 29-row head (three answers + marks), class-mean direction for unsure, per-mode temperatures

mihailgribov
22804364d ago

Qwen3.5-4B cut at block 23, three-row head

mihailgribov
59c16de4d ago

initial commit

mihailgribov