Niko1221/Strata
1
0 Fork 0
mirror of https://github.com/Niko1221/Strata.git synced 2026-11-07 14:21:06 -10:00
Code Issues Packages Projects Releases 1 Wiki Activity

DeepSeek: IST-DAS 1.75 bpw (IQ1_M), the smallest official set, as model ist-das-1.75bpw: pins at ISTA-DASLab cb02422c (the 9 earlier files are unchanged there), fit numbers, setup recommends it where 2.00 is tight in memory or disk (warn, never refuse), catalog/TUI text, docs, tests

Browse Source
This commit is contained in:
Niko1221 committed 2026-10-24 02:08:43 +02:00
1 parent eec705cc98
commit c5e0e14823
7 files changed
+118 -23
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

docs/DEEPSEEK41.md
+4 -2
View File
@@ -26,15 +26,16 @@ with an out-of-memory message. Both are said before anything is downloaded.
## The official 2-bit versions (the ones for a normal PC)
Setup offers two official sets (ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF, pinned): **IST-DAS 2.00 bpw** (138 GB) and **IST-DAS 2.25 bpw** (155 GB). Both use the
Setup offers three official sets (ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF, pinned): **IST-DAS 2.00 bpw** (138 GB), **IST-DAS 2.25 bpw** (155 GB) and the smallest, **IST-DAS 1.75 bpw** (121 GB, slightly lower quality; setup recommends it where 2.00 is tight in memory or disk). All use the
same 111 GB n-gram table (downloaded once into `models/deepseek-v4.1-flash-shared`) and the 8 GB DSpark drafter: 257 GB of disk for
2.00, 274 GB for 2.25, 412 GB for both. Quality against the original model: perplexity +16% on 2.00, +8% on 2.25 (code and agent work
2.00, 274 GB for 2.25, 239 GB for 1.75, 412 GB for 2.00 and 2.25 together. Quality against the original model: perplexity +16% on 2.00, +8% on 2.25 (code and agent work
about +3.5% / +2.5% on 2.25).
| File | Size | What it is |
| --- | ---: | --- |
| IQ2_XXS shards (2.00 bpw), 4 files | 138 GB | the model, first set |
| IQ2_XS shards (2.25 bpw), 4 files | 155 GB | the model, second set |
| IQ1_M shards (1.75 bpw), 4 files | 121 GB | the model, smallest set |
| IQ4_NL: `DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf` | 111 GB | the 4-bit Engram lookup table the model reads from RAM or SSD. Not a third model: both sets use this one file, and it is already in their download sizes |
| DSpark drafter (Q8_0) | 8 GB | the small draft model (Unsloth) |
@@ -56,6 +57,7 @@ and its 7 GB of experts take the graphics memory); the RAM store sizes itself. L
```
./setup.sh --family deepseek41 --model IST-DAS-2.00bpw # 257 GB: the official 2.00 set + Engram + drafter (resumable, SHA-256 checked)
./setup.sh --family deepseek41 --model IST-DAS-2.25bpw # 274 GB (the Engram is shared with 2.00)
./setup.sh --family deepseek41 --model IST-DAS-1.75bpw # 239 GB, the smallest
./setup.sh --family deepseek41 --model UD-Q8_K_XL # Unsloth's 508 GB set, for hundreds of GB of RAM or several cards
./setup.sh --family deepseek41 --gguf-dir DIR # files you already have: the shards (or the folder with 2.00bpw / 2.25bpw / UD-Q8_K_XL inside); the Engram and drafter are looked for beside them
strata launch --model ist-das-2.00bpw --plan # where it would go on this computer, nothing changed
setup.py
+22 -6
View File
@@ -99,7 +99,7 @@ HF_REVISIONS = {
"unsloth/Qwen3.6-35B-A3B-MTP-GGUF": "5bc3e238d916f48a861bac2f8a1990a0e9b7e98d", # 2026-10-12
"agentionai/Qwen3.8-Flash-Next-Gyro-GGUF": "1ada08422e404aca51ab9df665f089f34c419c6b", # 2026-10-13
"unsloth/DeepSeek-V4.1-Flash-1017-GGUF": "f924b94f19c1fea02c06a65c8475b0a45e2b88b1", # 2026-10-20 (the port's reference)
"ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF": "0d8f49fb15df73789a9c237a1b8b7b088423dbc0", # 2026-10-24 (ISTA-DASLab's final 2.00 / 2.25 bpw sets and Engram)
"ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF": "cb02422c900df489cea5833e8a92171224059b3c", # 2026-10-24 (ISTA-DASLab's final 1.75 / 2.00 / 2.25 bpw sets and Engram)
}
@@ -679,7 +679,7 @@ DSV41_FILES = {
"mmproj-BF16.gguf": (972208288, "9a1ef7c40d0d702c7ed8c5f863f60abae64091387ba3fce0ad19ce9beeda3bf9"),
}
# The two OFFICIAL IST-DAS sets (ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF at the revision above, a flat repository; Hub LFS sha256 = what the
# files hash to): IQ2_XXS is the 2.00 bpw set and IQ2_XS the 2.25 bpw set, 4 shards each, plus ONE Engram table (110.6 GB, the same file for both sets:
# files hash to): IQ2_XXS is the 2.00 bpw set, IQ2_XS the 2.25 bpw set and IQ1_M the 1.75 bpw set, 4 shards each, plus ONE Engram table (110.6 GB, the same file for both sets:
# downloaded once into the shared folder) and the DSpark drafter and the 5.8 MB metadata shard of Unsloth's set that pack_gguf --engram-meta reads
# (both pinned in DSV41_FILES above). This is the set a normal PC runs (64 GB RAM + a 12 GB card; docs/DEEPSEEK41.md); the UD-Q8_K_XL entry stays
# for the big machines.
@@ -688,6 +688,10 @@ DSV41_ISTA_ENGRAM = "DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf"
DSV41_ISTA_SHARD = "DeepSeek-V4.1-Flash-1017-GSQ-RCO-{quant}-{i:05d}-of-00004.gguf"
DSV41_DSPARK = "dspark-DeepSeek-V4.1-Flash-1017-Q8_0.gguf"
DSV41_ISTA_FILES = {
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ1_M-00001-of-00004.gguf": (39358830464, "6e3f0d5ff53d00428ec32c079db9e8255cb410ccaa435e5342aaa8a47fe1339d"),
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ1_M-00002-of-00004.gguf": (39176113248, "34d68c99fdae9948f9f0af0dafe308e805ce0a78d2c48641b24cc57242fdc0af"),
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ1_M-00003-of-00004.gguf": (39997334720, "a08468c83955e0c4d3a9eb3f7b86c45b7bbac55d99c4640039a296a8a4b327ea"),
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ1_M-00004-of-00004.gguf": (2188841824, "b05f26a92325aa6845cebcb3ec1f93d2e65ebf89bdc3f805695960d2f0681832"),
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XS-00001-of-00004.gguf": (39535776992, "4c65307992ab586c527ca1f6f64062aa96a33b23c02b7fed103f45f75e854798"),
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XS-00002-of-00004.gguf": (39494617536, "933461a974fcccee741bf2f9798201e72ab4aa3f8a952be56f849896ca6ea30e"),
"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XS-00003-of-00004.gguf": (39423838496, "bbf91fa346a85ebaf32dd4fd5c456b534d780d381e54f46ac90783a481373240"),
@@ -757,6 +761,13 @@ MULTI_FILE_FAMILIES["deepseek41"]["models"] = {
"ram_gb": 44, "recipe": "ista", "subdir": None, "legacy_subdir": "2.25bpw", "profile": "profile-2.25.bin", "hf": hf(DSV41_ISTA_REPO),
"shards": [DSV41_ISTA_SHARD.format(quant="IQ2_XS", i=i) for i in range(1, 5)],
"shared": [DSV41_ISTA_ENGRAM, DSV41_SHARD.format(i=1), DSV41_DSPARK]},
"IST-DAS-1.75bpw": {
"label": "IST-DAS 1.75 bpw",
"about": "the smallest official version: 121 GB of model + 111 GB n-gram table + 8 GB drafter (239 GB of disk, 18 GB less than 2.00). Runs "
"where 2.00 is tight (a 12 GB card and 32-40 GB of RAM); slightly lower quality than 2.00",
"ram_gb": 34, "recipe": "ista", "subdir": None, "legacy_subdir": None, "profile": "profile-2.00.bin", "hf": hf(DSV41_ISTA_REPO),
"shards": [DSV41_ISTA_SHARD.format(quant="IQ1_M", i=i) for i in range(1, 5)],
"shared": [DSV41_ISTA_ENGRAM, DSV41_SHARD.format(i=1), DSV41_DSPARK]},
"UD-Q8_K_XL": {
"label": "Unsloth UD-Q8_K_XL",
"about": "Unsloth Dynamic 8-bit: the experts in MXFP4 as DeepSeek made them, the rest in BF16 (the checkpoint's own precision); "
@@ -7917,9 +7928,14 @@ def install_dsv41(a, family, gpu, hip, amd, cuda_tk, multi, port, ram, adopted,
if a.model:
model = a.model
else:
try:
room = free_gb(Path(a.models_dir) if Path(a.models_dir).exists() else Path(a.models_dir).parent)
except Exception: # noqa: BLE001 - the disk is only a hint here
room = None
rec = dsv41_fit.recommend(ram, room) # the best version this computer's memory and disk suit; warn-never-refuse
for i, m in enumerate(names, 1):
say(f" {i}) {fam['models'][m].get('label', m)}: {fam['models'][m]['about']}" + (" (recommended)" if i == 1 else ""))
model = names[int(ask("Which version?", [str(i) for i in range(1, len(names) + 1)], "1", a.yes)) - 1]
say(f" {i}) {fam['models'][m].get('label', m)}: {fam['models'][m]['about']}" + (" (recommended)" if m == rec else ""))
model = names[int(ask("Which version?", [str(i) for i in range(1, len(names) + 1)], str(names.index(rec) + 1), a.yes)) - 1]
ok(f"size: {model}")
md = fam["models"][model]
say(" Its GGUF files: " + ("IST-DAS / ISTA-DASLab (huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF), with the DSpark drafter and a small metadata file from Unsloth"
@@ -8363,7 +8379,7 @@ def main() -> int:
ap.add_argument("--family", choices=list(FAMILIES) + list(PREVIEW_FAMILIES) + list(MULTI_FILE_FAMILIES),
help="qwen = Qwen3.8-Flash-Next, swift = Swift 1.5, qwen36 = Qwen3.6-35B-A3B (pictures with --vision, "
"several requests at once with --parallel; docs/QWEN36.md), deepseek41 = DeepSeek-V4.1-Flash "
"(the official 2.00 / 2.25 bpw versions, 257 / 274 GB; docs/DEEPSEEK41.md)")
"(the official 1.75 / 2.00 / 2.25 bpw versions, 239 / 257 / 274 GB; docs/DEEPSEEK41.md)")
ap.add_argument("--model", choices=list(dict.fromkeys([*MODELS, *(m for f in PREVIEW_FAMILIES.values()
for m in f["models"]),
*(m for f in MULTI_FILE_FAMILIES.values() for m in f["models"])])))
@@ -8830,7 +8846,7 @@ def main() -> int:
try: # DeepSeek-V4.1-Flash: a line, from the real numbers (strata_control/dsv41_fit.py)
from strata_control import dsv41_fit
cards_ok = amd_ok if hip else [g for g in found if gpu_problem(g) is None]
for mname in ("IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "UD-Q8_K_XL"):
for mname in ("IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "IST-DAS-1.75bpw", "UD-Q8_K_XL"):
ds = dsv41_fit.plan(cards_ok, ram, None, model=mname)
tail = {"fits": f"fits ({ds['experts_in_vram']:,} of {ds['experts_total']:,} experts in graphics memory, the rest on the CPU)",
"tight": ("runs, with the experts RAM cannot hold read from the SSD (about 5-7 tokens a second measured)"
strata_control/dsv41_fit.py
+21 -2
View File
@@ -61,6 +61,11 @@ ISTA_SETS = {
"IST-DAS-2.00bpw": {"title": "IST-DAS 2.00 bpw", "set_bytes": 138_238_893_056, "experts_gb": 132.07, "profile": "profile-2.00.bin",
"ppl": "16%", "ram_ok_gb": 40.0, "ram_great_gb": 160.0, "disk_total": 257.0,
"quality": "about 16% above the original model's perplexity (the 2-bit price)"},
# IQ1_M (1.75 bpw): 120,721,120,256 bytes of shards; experts = the shards minus the 6.16 GB of dense weights. RAM class from the experts' size against
# the 2.00 set's measured 40 GB (not measured on a computer yet); quality: placeholder until the H100 score is in.
"IST-DAS-1.75bpw": {"title": "IST-DAS 1.75 bpw", "set_bytes": 120_721_120_256, "experts_gb": 114.56, "profile": "profile-2.00.bin",
"ram_ok_gb": 34.0, "disk_total": 239.0,
"quality": "slightly lower than the 2.00 version (the smallest; a quality figure follows when it is measured)"},
"IST-DAS-2.25bpw": {"title": "IST-DAS 2.25 bpw", "set_bytes": 155_225_824_256, "experts_gb": 149.06, "profile": "profile-2.25.bin",
"ppl": "8%", "ram_ok_gb": 44.0, "ram_great_gb": 176.0, "disk_total": 274.0,
"quality": "about 8% above the original model's perplexity (code and agent work about 2.5-3.5% above), "
@@ -89,6 +94,20 @@ def ista_download_bytes(model: str, have_engram: bool = False) -> int:
return v["set_bytes"] + ISTA_DSPARK_BYTES + ISTA_META_BYTES + (0 if have_engram else ISTA_ENGRAM_BYTES)
def recommend(ram_gb: float | None, disk_free_gb: float | None = None) -> str:
"""The official set that suits this computer: 2.00 bpw, unless its memory (under the measured 40 GB of RAM) or its disk (under 2.00's need but
enough for the smaller one) make the 1.75 bpw set the better start. Only a recommendation: every set can still be chosen."""
big, small = "IST-DAS-2.00bpw", "IST-DAS-1.75bpw"
if ram_gb is not None and float(ram_gb) < ISTA_SETS[big]["ram_ok_gb"]:
return small
if disk_free_gb is not None:
need_big = ista_download_bytes(big) / GB + DISK_HEADROOM_GB
need_small = ista_download_bytes(small) / GB + DISK_HEADROOM_GB
if float(disk_free_gb) < need_big and float(disk_free_gb) >= need_small:
return small
return big
def _round1(x: float) -> float:
return round(float(x) * 10) / 10
@@ -153,7 +172,7 @@ def _plan_ista(model: str, cards: list[dict] | None, ram_gb: float | None, disk_
lines.append(f"DeepSeek-V4.1-Flash, {v['title']}: a {download:.0f} GB download ({set_gb:.0f} GB of model, {ISTA_ENGRAM_BYTES / GB:.0f} GB "
f"n-gram table, {ISTA_DSPARK_BYTES / GB:.0f} GB drafter). Quality: {v['quality']}.")
lines.append("The n-gram table stays on the SSD and is read a few rows for each word: use an NVMe drive for the model files (a slow "
"disk makes every word slower). The same table serves both IST-DAS sets, so installing the second one downloads it once.")
"disk makes every word slower). The same table serves every IST-DAS set, so installing another one downloads it once.")
if n == 0:
out["verdict"] = "no_card"
warns.append("No graphics card Strata can use was found: this model needs one or more cards (8 GB or more; 12 GB is what it was "
@@ -219,7 +238,7 @@ def plan(cards: list[dict] | None, ram_gb: float | None, disk_free_gb: float | N
vendor: str | None = None, model: str | None = None) -> dict[str, Any]:
"""Where the model goes on this computer. `cards`: the graphics cards ({"index", "name", "vram_gb", ...}, as gpu_fit takes them),
`ram_gb` the RAM, `disk_free_gb` the free space where the files go (None: not checked). `model`: an official IST-DAS set
(ISTA_SETS: "IST-DAS-2.00bpw" / "IST-DAS-2.25bpw") gets its own numbers; None or "UD-Q8_K_XL" is Unsloth's 508 GB set.
(ISTA_SETS: "IST-DAS-2.00bpw" / "IST-DAS-2.25bpw" / "IST-DAS-1.75bpw") gets its own numbers; None or "UD-Q8_K_XL" is Unsloth's 508 GB set.
Keys: cards (per card: index, name, vram_gb, layers, dense_gb, experts_gb, experts), experts_in_vram, experts_total, vram_experts_gb,
cpu_experts_gb, ram_needed_gb, spill_gb, spare_ram_gb, verdict (fits / tight / spills / unknown / no_card), tp, args,
tests/terminal/test_launch.py
+1 -1
View File
@@ -110,7 +110,7 @@ class CatalogRules(unittest.TestCase):
self.assertEqual(source, "setup.py") # the same tables setup.py and the app read
for r in rows:
if r.family == "deepseek41": # its sizes are in MULTI_FILE_FAMILIES (shards + shared files), not in MODELS
self.assertAlmostEqual(r.download_gb, {"UD-Q8_K_XL": 508.4, "IST-DAS-2.00bpw": 256.8, "IST-DAS-2.25bpw": 273.8}[r.model], delta=0.1)
self.assertAlmostEqual(r.download_gb, {"UD-Q8_K_XL": 508.4, "IST-DAS-2.00bpw": 256.8, "IST-DAS-2.25bpw": 273.8, "IST-DAS-1.75bpw": 239.3}[r.model], delta=0.1)
continue
self.assertAlmostEqual(r.download_gb, models[r.model]["download_gb"])
rec = [r for r in rows if r.recommended]
tests/terminal/test_launch_dsv41.py
+3 -3
View File
@@ -36,10 +36,10 @@ class LaunchDeepSeek(unittest.TestCase):
def test_the_model_is_in_the_list(self):
rows = [r for r in catalog(self.hw).rows() if r.family == "deepseek41"]
self.assertEqual([r.model for r in rows], ["IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "UD-Q8_K_XL"])
self.assertEqual([r.download_gb for r in rows[:2]], [256.8, 273.8])
self.assertEqual([r.model for r in rows], ["IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "IST-DAS-1.75bpw", "UD-Q8_K_XL"])
self.assertEqual([r.download_gb for r in rows[:3]], [256.8, 273.8, 239.3])
self.assertEqual(catalog(self.hw).resolve("ist-das-2.25bpw"), ("deepseek41", "IST-DAS-2.25bpw"))
r = rows[2]
r = rows[3]
self.assertEqual((r.id, r.download_gb, r.verdict, r.images), ("deepseek-v4.1-flash-ud-q8_k_xl", 508.4, "fits", False))
self.assertEqual(catalog(self.hw).resolve("ud-q8_k_xl"), ("deepseek41", "UD-Q8_K_XL"))
self.assertEqual(catalog(self.hw).max_context("deepseek41", "UD-Q8_K_XL"), 8192)
tools/strata_mcp.py
+2 -1
View File
@@ -1290,7 +1290,8 @@ class Strata:
if pl.get("variant"): # an official IST-DAS set (strata_control/dsv41_fit.py): measured on 46-64 GB, never a refusal
if v == "spills":
return (f"slow here: this computer has {pl['ram_gb']:.0f} GB of RAM and the set was measured with 46-64 GB, so most of its "
"experts are read from the SSD; it can still be installed")
"experts are read from the SSD; it can still be installed"
+ ("" if pl["variant"] == "IST-DAS-1.75bpw" else " (the smaller 1.75 bpw version suits this computer better)"))
if v == "tight":
return ("tight: runs, with the experts RAM cannot hold read from the SSD (about 5-7 tokens a second measured on 46-64 GB "
"of RAM and a 12 GB card)")
tools/test_setup_dsv41.py
+65 -8
View File
@@ -288,7 +288,7 @@ class Catalog(unittest.TestCase):
s = M.Strata(ROOT)
fam = next(f for f in s.model_table({"ram_gb": 251.0, "gpus": [{**c, "usable": True} for c in FOUR], "engine_vendor": "amd"})
if f["family"] == "deepseek41")
self.assertEqual([x["model"] for x in fam["sizes"]], ["IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "UD-Q8_K_XL"])
self.assertEqual([x["model"] for x in fam["sizes"]], ["IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "IST-DAS-1.75bpw", "UD-Q8_K_XL"])
small_sets = {x["model"]: x for x in fam["sizes"]}
self.assertEqual((small_sets["IST-DAS-2.00bpw"]["download_gb"], small_sets["IST-DAS-2.25bpw"]["download_gb"]), (256.8, 273.8))
self.assertTrue(small_sets["IST-DAS-2.00bpw"]["on_this_pc"].startswith("fits"))
@@ -326,21 +326,22 @@ class OfficialSets(unittest.TestCase):
def test_the_pinned_files(self):
fam = setup.MULTI_FILE_FAMILIES["deepseek41"]
self.assertEqual(list(fam["models"]), ["IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "UD-Q8_K_XL"]) # the official sets first
self.assertEqual(list(fam["models"]), ["IST-DAS-2.00bpw", "IST-DAS-2.25bpw", "IST-DAS-1.75bpw", "UD-Q8_K_XL"]) # the official sets first
repo = "ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF"
self.assertEqual(setup.HF_REVISIONS[repo], "0d8f49fb15df73789a9c237a1b8b7b088423dbc0")
self.assertEqual(setup.HF_REVISIONS[repo], "cb02422c900df489cea5833e8a92171224059b3c")
self.assertIs(setup.REPO_PINS[repo], setup.DSV41_ISTA_FILES)
self.assertNotIn("helcig/tmp-oct26-2", setup.HF_REVISIONS) # the first release's repository is only remembered as local names
self.assertEqual(setup.DSV41_ISTA_REPO, repo)
self.assertEqual(len(setup.DSV41_ISTA_FILES), 9) # 2 x 4 shards + the Engram
self.assertEqual(len(setup.DSV41_ISTA_FILES), 13) # 3 x 4 shards + the Engram
for n, (size, sha) in setup.DSV41_ISTA_FILES.items():
self.assertRegex(sha, r"^[0-9a-f]{64}$", n)
eng = setup.DSV41_ISTA_FILES["DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf"]
self.assertEqual(eng, (110_595_290_816, "eb19f30d368d393784e1fc54fdfaffc427ebb2110bf3762fdf5a8fbb9a2d740a"))
self.assertEqual(setup.DSV41_ALL_FILES["dspark-DeepSeek-V4.1-Flash-1017-Q8_0.gguf"],
(7_976_914_240, "6ba6ac1b331c540691a209e7fc97dec2a24e295acb1e755c84c7d24a4373927a"))
sums = {m: sum(setup.DSV41_ALL_FILES[n][0] for n in fam["models"][m]["shards"]) for m in list(fam["models"])[:2]}
self.assertEqual(sums, {"IST-DAS-2.00bpw": 138238893056, "IST-DAS-2.25bpw": 155225824256})
sums = {m: sum(setup.DSV41_ALL_FILES[n][0] for n in fam["models"][m]["shards"]) for m in list(fam["models"])[:3]}
self.assertEqual(sums, {"IST-DAS-2.00bpw": 138238893056, "IST-DAS-2.25bpw": 155225824256, "IST-DAS-1.75bpw": 120721120256})
self.assertEqual(sums["IST-DAS-1.75bpw"], dsv41_fit.ISTA_SETS["IST-DAS-1.75bpw"]["set_bytes"])
self.assertEqual(sums["IST-DAS-2.00bpw"], dsv41_fit.ISTA_SETS["IST-DAS-2.00bpw"]["set_bytes"])
self.assertEqual(sums["IST-DAS-2.25bpw"], dsv41_fit.ISTA_SETS["IST-DAS-2.25bpw"]["set_bytes"])
self.assertEqual(setup.DSV41_ISTA_FILES["DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf"][0], dsv41_fit.ISTA_ENGRAM_BYTES)
@@ -349,8 +350,8 @@ class OfficialSets(unittest.TestCase):
for n in fam["models"][m]["shared"]:
self.assertIn(n, setup.DSV41_SHARED)
self.assertIn(n, setup.DSV41_ALL_FILES)
self.assertRegex(setup.DSV41_SHARED[setup.DSV41_ISTA_ENGRAM], r"ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/0d8f49fb[0-9a-f]{32}/$")
self.assertRegex(fam["models"]["IST-DAS-2.25bpw"]["hf"], r"ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/0d8f49fb[0-9a-f]{32}/$")
self.assertRegex(setup.DSV41_SHARED[setup.DSV41_ISTA_ENGRAM], r"ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/cb02422c[0-9a-f]{32}/$")
self.assertRegex(fam["models"]["IST-DAS-2.25bpw"]["hf"], r"ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/cb02422c[0-9a-f]{32}/$")
def test_shipped_profiles_match_their_hashes(self):
import hashlib
@@ -468,6 +469,49 @@ class LegacyFiles(unittest.TestCase):
self.assertEqual(got, old)
class Smallest(unittest.TestCase):
"""IST-DAS 1.75 bpw (IQ1_M): the smallest official set, recommended where 2.00 is tight, never forced."""
def test_the_files_the_pins_and_the_recipe(self):
fam = setup.MULTI_FILE_FAMILIES["deepseek41"]
d = fam["models"]["IST-DAS-1.75bpw"]
self.assertEqual(d["recipe"], "ista")
self.assertEqual(d["shards"], [f"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ1_M-{i:05d}-of-00004.gguf" for i in range(1, 5)])
self.assertEqual(d["shared"], fam["models"]["IST-DAS-2.00bpw"]["shared"]) # the same Engram, drafter and metadata shard
self.assertEqual(setup.DSV41_ISTA_FILES[d["shards"][0]],
(39358830464, "6e3f0d5ff53d00428ec32c079db9e8255cb410ccaa435e5342aaa8a47fe1339d"))
self.assertEqual(setup.DSV41_ISTA_FILES[d["shards"][3]][0], 2188841824)
self.assertEqual(sum(setup.DSV41_ISTA_FILES[n][0] for n in d["shards"]), 120_721_120_256)
self.assertTrue(all(n not in setup.DSV41_LEGACY_NAMES.values() for n in d["shards"])) # no first-release names for it
cmd = setup.dsv41_pack_cmd(Path("/m"), d["shards"], Path("/s/e.gguf"), Path("/s/m.gguf"), Path("/s/d.gguf"), Path("/p"))
self.assertIn("--dspark-noise-token", cmd)
self.assertNotIn("--skip-tensors", cmd)
def test_it_is_the_smallest_and_warns_never_refuses(self):
sizes = {m: dsv41_fit.ista_download_bytes(m) / 1e9 for m in dsv41_fit.ISTA_SETS}
self.assertEqual(min(sizes, key=sizes.get), "IST-DAS-1.75bpw")
self.assertAlmostEqual(sizes["IST-DAS-1.75bpw"], 239.3, delta=0.1)
for ram in (24.0, 36.0, 64.0):
p = dsv41_fit.plan([RTX5070], ram, None, model="IST-DAS-1.75bpw")
self.assertIn(p["verdict"], ("fits", "tight", "spills"))
self.assertTrue(p["lines"])
self.assertEqual(dsv41_fit.plan([RTX5070], 36.0, None, model="IST-DAS-1.75bpw")["verdict"], "tight") # runs where 2.00 spills
self.assertEqual(dsv41_fit.plan([RTX5070], 36.0, None, model="IST-DAS-2.00bpw")["verdict"], "spills")
def test_it_is_recommended_where_2_00_is_tight(self):
self.assertEqual(dsv41_fit.recommend(64.0, 900.0), "IST-DAS-2.00bpw")
self.assertEqual(dsv41_fit.recommend(None, None), "IST-DAS-2.00bpw") # unknown: the usual one
self.assertEqual(dsv41_fit.recommend(32.0, 900.0), "IST-DAS-1.75bpw") # less RAM than 2.00 was measured with
self.assertEqual(dsv41_fit.recommend(64.0, 250.0), "IST-DAS-1.75bpw") # disk fits 1.75 (239 + 6) but not 2.00 (257 + 6)
self.assertEqual(dsv41_fit.recommend(64.0, 100.0), "IST-DAS-2.00bpw") # fits neither: no reason to prefer it
def test_the_menu_marks_the_recommendation_and_still_lets_anyone_choose(self):
import io
names = list(setup.MULTI_FILE_FAMILIES["deepseek41"]["models"])
self.assertIn("IST-DAS-1.75bpw", names)
self.assertEqual(names.index(dsv41_fit.recommend(32.0, 900.0)) + 1, 3)
class InstallOfficial(Install):
"""install_dsv41 for an official set, the engine and the tools mocked."""
@@ -581,5 +625,18 @@ class InstallOfficial(Install):
self.assertIn("not enough free disk space", out)
class InstallSmallest(InstallOfficial):
def test_downloads_the_iq1_m_files_and_the_shared_ones(self):
rc, out = self.install_ista(model="IST-DAS-1.75bpw", ram=36.0)
self.assertEqual(rc, 0, out)
get = [c for c in self.calls if c[0] == "get"]
self.assertEqual([c[2].name for c in get][:4], [f"DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ1_M-{i:05d}-of-00004.gguf" for i in range(1, 5)])
self.assertTrue(get[0][1].startswith(setup.hf(setup.DSV41_ISTA_REPO)))
self.assertIn("cb02422c900df489cea5833e8a92171224059b3c", get[0][1])
self.assertIn("239 GB", out)
a = self.cfg["args"]
self.assertEqual(a[a.index("--dspark") + 1], "auto")
if __name__ == "__main__":
unittest.main()