vh/esh-pfi-infrastructure
1
0
Fork 0
Code Issues 7 Pull Requests Actions Packages Projects Releases Wiki Activity

feat(intern-decision): stack, DNS and GPU 3 acceptance for the SemIf replacement

Browse Source 
stacks/intern-decision: compose (GPU 1, :8033, hard VRAM cap as the single .env knob,
healthcheck, Homepage group 'AI - Eval & Retrieval'), .env.example and README.
dns: intern-decision.fv.internal -> fv-ml1 (synced to ana/esh/nh3).
acceptance on fv-ml1 GPU 3, 3 fresh processes: bit-identical to the Jev bench's native rows
(pooled 240/259, Wyrd 79/84, 0/560 flips, Δp 0), negative control 10/122/14, 0 flips across
restarts; largest accepted request 200 at a 10,134 MiB card peak under a 9.25 GiB cap; 503 and
recovery proven at a tight cap. GPU 1 deploy held: nvidia-smi Free on GPU 1 is 15,442 MiB.
This commit is contained in:
vh
2026-09-30 09:38:00 -07:00
parent 21d16d7ad8
commit a262477a61
59 changed files with 11969 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
services/intern-decision-serve/acceptance/gpu3-2026-09-30/uncapped/checks.json
+357
View File
@@ -0,0 +1,357 @@
{
"url": "http://127.0.0.1:18033",
"started_utc": "2026-09-30T16:09:34Z",
"health": {
"status": "ok",
"model": {
"name": "Intern-Decision-4B",
"source": "internlm/Intern-Decision-4B",
"revision": "0e5e6aa7d6d750e2b1504ba11a8136cb58aeb3cd",
"checkpoint": "/hf/hub/models--internlm--Intern-Decision-4B/snapshots/0e5e6aa7d6d750e2b1504ba11a8136cb58aeb3cd",
"inference_py_sha256": "c904e2c67ca0775621a22375ee373d2ba30b52117cda870c6c9ef74143b29863",
"temperature": 1.99241824,
"dtype": "bfloat16",
"attn_implementation": "sdpa",
"device": "cuda",
"max_length": 8192,
"torch_version": "2.10.0+cu128",
"transformers_version": "5.17.0",
"vision_tower": "removed",
"device_name": "NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition",
"allocated_gib": 7.937,
"reserved_gib": 7.969,
"max_reserved_gib": 8.861
},
"vram_cap_gib": null,
"max_tokens": 8192,
"max_decisions": 64,
"max_questions_per_call": 16,
"chunking": "/decide/shared questions are packed greedily, in request order, into calls of at most 16 (1-16, 17-32, ...); each call is one prompt, so the questions in a call are asked together. With orderings, ordering k of every decision forms wave k, packed the same way.",
"workloads": []
},
"auth": {
"/decide": {
"no_token": 401,
"wrong_token": 401,
"right_token": 200
},
"/decide/shared": {
"no_token": 401,
"wrong_token": 401,
"right_token": 200
},
"health_no_token": 200,
"pass": true,
"t_start": 1790784574.0391982,
"t_end": 1790784574.1316206
},
"chunking": {
"status": [
200,
200,
200
],
"timing": {
"total_seconds": 0.14785360009409487,
"batch_size": 20,
"calls": 2,
"questions_per_call": [
16,
4
],
"input_tokens": [
1884,
1328
],
"inference_seconds": 0.13485
},
"rows": [
{
"id": "d0",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q1",
"questions": 16
},
"sha_equal": true
},
{
"id": "d1",
"top_whole": "no",
"top_split": "no",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q2",
"questions": 16
},
"sha_equal": true
},
{
"id": "d2",
"top_whole": "no",
"top_split": "no",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q3",
"questions": 16
},
"sha_equal": true
},
{
"id": "d3",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q4",
"questions": 16
},
"sha_equal": true
},
{
"id": "d4",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q5",
"questions": 16
},
"sha_equal": true
},
{
"id": "d5",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q6",
"questions": 16
},
"sha_equal": true
},
{
"id": "d6",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q7",
"questions": 16
},
"sha_equal": true
},
{
"id": "d7",
"top_whole": "no",
"top_split": "no",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q8",
"questions": 16
},
"sha_equal": true
},
{
"id": "d8",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q9",
"questions": 16
},
"sha_equal": true
},
{
"id": "d9",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q10",
"questions": 16
},
"sha_equal": true
},
{
"id": "d10",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q11",
"questions": 16
},
"sha_equal": true
},
{
"id": "d11",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q12",
"questions": 16
},
"sha_equal": true
},
{
"id": "d12",
"top_whole": "no",
"top_split": "no",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q13",
"questions": 16
},
"sha_equal": true
},
{
"id": "d13",
"top_whole": "no",
"top_split": "no",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q14",
"questions": 16
},
"sha_equal": true
},
{
"id": "d14",
"top_whole": "no",
"top_split": "no",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q15",
"questions": 16
},
"sha_equal": true
},
{
"id": "d15",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 0,
"field": "q16",
"questions": 16
},
"sha_equal": true
},
{
"id": "d16",
"top_whole": "unclear",
"top_split": "unclear",
"max_dp": 0.0,
"call_whole": {
"index": 1,
"field": "q1",
"questions": 4
},
"sha_equal": true
},
{
"id": "d17",
"top_whole": "unclear",
"top_split": "unclear",
"max_dp": 0.0,
"call_whole": {
"index": 1,
"field": "q2",
"questions": 4
},
"sha_equal": true
},
{
"id": "d18",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 1,
"field": "q3",
"questions": 4
},
"sha_equal": true
},
{
"id": "d19",
"top_whole": "yes",
"top_split": "yes",
"max_dp": 0.0,
"call_whole": {
"index": 1,
"field": "q4",
"questions": 4
},
"sha_equal": true
}
],
"e2e_ms": 151.3,
"pass": true,
"t_start": 1790784574.1316838,
"t_end": 1790784574.4364655
},
"maxreq": {
"state_chars": 11337,
"tokens_per_call": 8192,
"decisions": 64,
"options_each": 16,
"body_bytes": 100953,
"runs": [
{
"status": 422,
"e2e_ms": 415.8,
"code": "invalid_request",
"calls": null,
"input_tokens": null,
"reserved_gib_after": 7.969,
"max_reserved_gib": 9.557,
"t_end": 1790784583.8459144
},
{
"status": 422,
"e2e_ms": 405.2,
"code": "invalid_request",
"calls": null,
"input_tokens": null,
"reserved_gib_after": 7.969,
"max_reserved_gib": 9.557,
"t_end": 1790784584.2551143
},
{
"status": 422,
"e2e_ms": 406.5,
"code": "invalid_request",
"calls": null,
"input_tokens": null,
"reserved_gib_after": 7.969,
"max_reserved_gib": 9.557,
"t_end": 1790784584.6652017
}
],
"t_start": 1790784574.4365265,
"t_end": 1790784584.6653428
},
"finished_utc": "2026-09-30T16:09:44Z"
}
services/intern-decision-serve/acceptance/gpu3-2026-09-30/uncapped/shape.json
+1
View File
@@ -0,0 +1 @@
{"label": "uncapped", "backend": "semif", "runs": 1, "per": 10, "max_questions": 10000, "events": [["start", 1790784611.4584568], ["short1:measure", 1790784611.5543973], ["crit21:measure", 1790784612.0878277], ["long1:measure", 1790784619.396992], ["long16:measure", 1790784627.8208334], ["shapes:done", 1790784629.9486265], ["end", 1790784629.9486306]], "shapes": {"short1": {"rows": 1, "runs": [{"e2e_ms": [35.92049889266491, 35.901617957279086, 35.708046052604914, 35.906028002500534, 36.10570100136101, 36.089200992137194, 35.76449700631201, 36.227591102942824, 35.77947709709406, 35.997220082208514], "server_ms": [34.09949690103531, 34.24146887846291, 34.126657992601395, 34.240649081766605, 34.342060098424554, 34.322259947657585, 34.07772700302303, 34.49505195021629, 34.051917027682066, 34.29409000091255], "statuses": [200, 200, 200, 200, 200, 200, 200, 200, 200, 200], "tokens": 181, "median_e2e": 35.91326344758272, "median_server": 34.24105898011476}], "tokens": 181, "median_e2e_ms": 35.9, "run_medians_e2e_ms": [35.9], "median_server_ms": 34.2, "non_200": 0}, "crit21": {"rows": 21, "runs": [{"e2e_ms": [85.90914588421583, 85.42978088371456, 85.99224709905684, 85.41093999519944, 85.5076820589602, 85.47824202105403, 85.63234284520149, 85.61140298843384, 85.56949300691485, 85.53116186521947], "server_ms": [79.33550095185637, 78.80261586979032, 79.46493290364742, 78.93323712050915, 78.8946459069848, 78.88813596218824, 79.00104694999754, 79.05412814579904, 79.03093798086047, 79.03374801389873], "statuses": [200, 200, 200, 200, 200, 200, 200, 200, 200, 200], "tokens": null, "median_e2e": 85.55032743606716, "median_server": 79.01599246542901}], "tokens": null, "median_e2e_ms": 85.6, "run_medians_e2e_ms": [85.6], "median_server_ms": 79.0, "non_200": 0}, "long1": {"rows": 1, "runs": [{"e2e_ms": [183.08503087610006, 183.89351991936564, 184.35147497802973, 185.54342817515135, 191.41989503987134, 183.4883950650692, 185.06936286576092, 185.85462193004787, 187.36803904175758, 184.89661114290357], "server_ms": [179.3815779965371, 180.2897888701409, 180.69604295305908, 181.9530779030174, 187.8542739432305, 179.9399449955672, 181.47349287755787, 182.3184520471841, 183.79759904928505, 181.30912980996072], "statuses": [200, 200, 200, 200, 200, 200, 200, 200, 200, 200], "tokens": 4019, "median_e2e": 184.98298700433224, "median_server": 181.3913113437593}], "tokens": 4019, "median_e2e_ms": 185.0, "run_medians_e2e_ms": [185.0], "median_server_ms": 181.4, "non_200": 0}, "long16": {"rows": 16, "runs": [{"e2e_ms": [210.3101098909974, 210.82147606648505, 212.6600358169526, 211.06627909466624, 214.81614094227552, 210.01119702123106, 215.28325602412224, 212.87801885046065, 211.02875797078013, 213.75816804356873], "server_ms": [203.99634796194732, 204.5704941265285, 206.43203612416983, 204.81025706976652, 208.57751998119056, 203.7482550367713, 208.42881803400815, 205.99671080708504, 204.55457409843802, 207.47225801460445], "statuses": [200, 200, 200, 200, 200, 200, 200, 200, 200, 200], "tokens": null, "median_e2e": 211.86315745580941, "median_server": 205.40348393842578}], "tokens": null, "median_e2e_ms": 211.9, "run_medians_e2e_ms": [211.9], "median_server_ms": 205.4, "non_200": 0}}}
services/intern-decision-serve/acceptance/gpu3-2026-09-30/uncapped/smi-gpu3-uncapped.csv
+1640
View File
File diff suppressed because it is too large Load Diff