pod backup 2026-11-07: e16 (0-158)

e16/S1-ABCLODO16-algebra2005/DONE

@@ -0,0 +1 @@
+2026-10-12 23:41:43

e16/S1-ABCLODO16-assist09/DONE

@@ -0,0 +1 @@
+2026-10-12 23:05:51

e16/S1-ABCLODO16-assist2017/DONE

@@ -0,0 +1 @@
+2026-10-13 12:01:45

e16/S1-ABCLODO16-dbe_kt22/DONE

@@ -0,0 +1 @@
+2026-10-13 15:34:03

e16/S1-ABCLODO16-junyi/DONE

@@ -0,0 +1 @@
+2026-10-13 06:05:49

e16/S1-ABCLODO16-nips34/DONE

@@ -0,0 +1 @@
+2026-10-13 04:38:38

e16/S1-ABCLODO16-xes3g5m/DONE

@@ -0,0 +1 @@
+2026-10-13 10:09:35

e16/S1-E19-a1-n16k-1model-syn/DONE

@@ -0,0 +1 @@
+2026-10-16 20:24:31

e16/S1-E19-a1-n16k-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-17 13:29:17

e16/S1-E19-a1-n16k-nopersona-syn/DONE

@@ -0,0 +1 @@
+2026-10-17 02:03:50

e16/S1-E19-a1-n16k-soft-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-16 14:49:17

e16/S1-E19-a1-n4k-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-15 23:45:40

e16/S1-E19-a2-n16k-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-21 19:12:15

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/adapter/README.md

@@ -0,0 +1,206 @@
+---
+base_model: /workspace/jev_kt/models/hf/JEV-9B
+library_name: peft
+tags:
+- base_model:adapter:/workspace/jev_kt/models/hf/JEV-9B
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.21.2

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/adapter/adapter_config.json

@@ -0,0 +1,57 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3_5ForCausalLM",
+    "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
+  },
+  "base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "kasa_config": null,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.21.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "v_proj",
+    "in_proj_z",
+    "up_proj",
+    "down_proj",
+    "gate_proj",
+    "o_proj",
+    "q_proj",
+    "out_proj",
+    "in_proj_qkv"
+  ],
+  "target_parameters": null,
+  "task_type": null,
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/adapter/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ee82a9f3632fc7d4ea37c7342ed4882562a01f0de796eaf07358116732a42be
+size 160486456

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/adapter_vllm/adapter_config.json

@@ -0,0 +1,58 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3_5ForCausalLM",
+    "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
+  },
+  "base_model_name_or_path": null,
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "kasa_config": null,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64.0,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.21.0",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "in_proj_qkv",
+    "in_proj_z",
+    "k_proj",
+    "lm_head",
+    "o_proj",
+    "out_proj",
+    "q_proj",
+    "up_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": null,
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:279e8a6fb1ccec91e68d622a5ab49ee8cacc2e22ecc5513b8988a21451926252
+size 176641832

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/adapter_vllm/decision_head.json

@@ -0,0 +1,99 @@
+{
+ "bias": [
+  0.0011035553179681301,
+  -0.0011038156226277351,
+  -0.0019555583130568266,
+  -0.0017452309839427471,
+  -0.0006223739474080503,
+  0.001689566415734589,
+  0.0015410069609060884,
+  -0.0021176303271204233,
+  -0.00041196096572093666,
+  -8.72826058184728e-05,
+  -3.86911015084479e-05,
+  0.000997072085738182,
+  -0.0019997144117951393,
+  -0.004234509542584419,
+  -0.004682670813053846,
+  -0.00647096848115325,
+  -0.006655825302004814,
+  -0.005742847453802824,
+  -0.006174066103994846,
+  -0.006459598895162344,
+  -0.005188263952732086,
+  -0.004888715222477913,
+  -0.0070209507830441,
+  -0.004854375030845404
+ ],
+ "verbalizer_ids": [
+  3721,
+  1802,
+  15,
+  16,
+  17,
+  18,
+  19,
+  20,
+  32,
+  33,
+  34,
+  35,
+  36,
+  37,
+  38,
+  39,
+  40,
+  41,
+  42,
+  43,
+  44,
+  45,
+  46,
+  47
+ ],
+ "slots": {
+  "num_slots": 24,
+  "ranges": {
+   "noul": [
+    0,
+    2
+   ],
+   "score": [
+    2,
+    8
+   ],
+   "choice": [
+    8,
+    24
+   ]
+  },
+  "verbalizers": [
+   "false",
+   "true",
+   "0",
+   "1",
+   "2",
+   "3",
+   "4",
+   "5",
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "template_version": "bare-v1"
+ },
+ "note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
+}

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/calibration.json

@@ -0,0 +1,29 @@
+{
+ "version": 1,
+ "per_kind": {
+  "noul": 1.2117223117345985,
+  "choice": 1.0711649063804969,
+  "score": 1.0221427041316569
+ },
+ "fit": {
+  "noul": {
+   "n": 1087,
+   "T": 1.2117223117345985,
+   "ce_before": 0.6236547827720642,
+   "ce_after": 0.6214077472686768
+  },
+  "choice": {
+   "n": 60,
+   "T": 1.0711649063804969,
+   "ce_before": 0.668276846408844,
+   "ce_after": 0.667122483253479
+  },
+  "score": {
+   "n": 53,
+   "T": 1.0221427041316569,
+   "ce_before": 1.0528051853179932,
+   "ce_after": 1.0526679754257202
+  }
+ },
+ "source": "models/e16/S1-E19-a2-n16k-a100-syn/ckpt-16000"
+}

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/head.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bd7d04f8a884aee1f2c998b38f23652f447ae0de0977283d36bd718743fb37f
+size 393464

e16/S1-E19-a2-n16k-a100-syn/ckpt-16000/judge_config.json

@@ -0,0 +1,164 @@
+{
+ "base_model_path": "/root/models/Qwen3.5-9B",
+ "hidden_size": 4096,
+ "slots": {
+  "num_slots": 24,
+  "ranges": {
+   "noul": [
+    0,
+    2
+   ],
+   "score": [
+    2,
+    8
+   ],
+   "choice": [
+    8,
+    24
+   ]
+  },
+  "verbalizers": [
+   "false",
+   "true",
+   "0",
+   "1",
+   "2",
+   "3",
+   "4",
+   "5",
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "template_version": "bare-v1"
+ },
+ "verbalizer_ids": [
+  3721,
+  1802,
+  15,
+  16,
+  17,
+  18,
+  19,
+  20,
+  32,
+  33,
+  34,
+  35,
+  36,
+  37,
+  38,
+  39,
+  40,
+  41,
+  42,
+  43,
+  44,
+  45,
+  46,
+  47
+ ],
+ "kinds": [
+  "noul",
+  "choice",
+  "score"
+ ],
+ "model_name": "jev-judge-qwen35-9b",
+ "model_version": "0.8.0",
+ "lm_head_included": true,
+ "weights_mode": "unmerged",
+ "adapter_subfolder": "adapter",
+ "lora": {
+  "r": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "target_modules": [
+   "down_proj",
+   "gate_proj",
+   "in_proj_qkv",
+   "in_proj_z",
+   "k_proj",
+   "o_proj",
+   "out_proj",
+   "q_proj",
+   "up_proj",
+   "v_proj"
+  ]
+ },
+ "trained_stage": "e16",
+ "source_checkpoint": "models/e16/S1-E19-a2-n16k-a100-syn/ckpt-16000",
+ "softcap": null,
+ "e16": {
+  "data": "data/e19/E19-a2-n16k-a100",
+  "data_manifest": {
+   "arm": "E19-a2-n16k-a100",
+   "name": null,
+   "train_sha256": "dd60c9d295faf488e4d339753884d09886018a3f703be161b45a25c5fce80988",
+   "n_train": 16000
+  },
+  "init": "released",
+  "objective": "s-log",
+  "sigma": 0.5,
+  "K": 8,
+  "reward": "brier",
+  "clip": 0.2,
+  "passes": 1,
+  "lr": 0.0001,
+  "lr_head": 0.0002,
+  "rows_per_step": 32,
+  "train_seed": 0,
+  "replay": "data/e16/replay/train.jsonl",
+  "replay_frac": 0.05,
+  "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
+  "max_len": 3072,
+  "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
+  "total_rows": 16000,
+  "lora": {
+   "r": 16,
+   "lora_alpha": 32,
+   "lora_dropout": 0.05
+  },
+  "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
+  "event": "ckpt",
+  "tag": "ckpt-16000",
+  "rows_seen": 16000,
+  "replay_seen": 842,
+  "step": 500,
+  "tokens": 8990682,
+  "tok_per_s": 801.8732506453285,
+  "rows_per_s": 1.5021273446902907,
+  "val": {
+   "n": 1200,
+   "dropped_long": 0,
+   "noul_ce": 0.6236547827720642,
+   "noul_T": 1.2117223117345985,
+   "choice_ce": 0.668276846408844,
+   "choice_T": 1.0711649063804969,
+   "score_ce": 1.0528051853179932,
+   "score_T": 1.0221427041316569,
+   "noul_auc": 0.682496,
+   "noul_acc": 0.648,
+   "noul_f1": 0.7021996615905245,
+   "noul_nll": 0.6334441490061581,
+   "noul_ece": 0.10943237079679967,
+   "noul_brier": 0.22197562435617027,
+   "noul_sharp": 0.1640363186597824,
+   "entropy": 0.6099246740341187
+  },
+  "dropped_long": 0,
+  "peak_gib": 18.235705852508545
+ }
+}

e16/S1-E19-a2-n16k-a100-syn/train_log.jsonl

@@ -0,0 +1,57 @@
+{"time": "16:05:21", "event": "start", "data": "data/e19/E19-a2-n16k-a100", "data_manifest": {"arm": "E19-a2-n16k-a100", "name": null, "train_sha256": "dd60c9d295faf488e4d339753884d09886018a3f703be161b45a25c5fce80988", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 95.56079339981079}
+{"time": "16:08:58", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 847.9972560047073, "loss": 0.5454432852566242, "entropy": 0.33670813776552677, "sharp": 0.38016169518232346}
+{"time": "16:12:34", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 849.9998772357029, "loss": 0.578414611518383, "entropy": 0.5799773186445236, "sharp": 0.21352986246347427}
+{"time": "16:15:57", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 847.736283145766, "loss": 0.3814765615388751, "entropy": 0.29598771780729294, "sharp": 0.40356129221618176}
+{"time": "16:19:25", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 847.6868112168953, "loss": 0.5857681110501289, "entropy": 0.6610568128526211, "sharp": 0.10525728762149811}
+{"time": "16:22:54", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 847.8339496057281, "loss": 0.394748168066144, "entropy": 0.3834071457386017, "sharp": 0.36977221071720123}
+{"time": "16:26:22", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 847.2997614429455, "loss": 0.4817565269768238, "entropy": 0.5584890320897102, "sharp": 0.22993380948901176}
+{"time": "16:29:52", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 847.5072536846753, "loss": 0.5975317303091288, "entropy": 0.35069425217807293, "sharp": 0.37037929706275463}
+{"time": "16:33:20", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 847.8746630331119, "loss": 0.5600134283304214, "entropy": 0.39232501201331615, "sharp": 0.35539371334016323}
+{"time": "16:36:51", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 847.8449371804938, "loss": 0.44657107442617416, "entropy": 0.4598126709461212, "sharp": 0.30851708352565765}
+{"time": "16:40:19", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 847.2081280685534, "loss": 0.4891582699492574, "entropy": 0.5046212431043386, "sharp": 0.2709479872137308}
+{"time": "16:41:48", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.6974984407424927, "noul_T": 2.329613324098613, "noul_auc": 0.6338120875597036, "noul_acc": 0.536, "noul_f1": 0.5303643724696356, "noul_nll": 0.6911487830877304, "noul_ece": 0.15971474444866182, "noul_brier": 0.24950373126953668, "noul_sharp": 0.10899211466312408, "entropy": 0.6582386493682861}
+{"time": "16:45:12", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 815.1333216005976, "loss": 0.585015382617712, "entropy": 0.4811478555202484, "sharp": 0.28365765884518623}
+{"time": "16:48:42", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 817.9071859750328, "loss": 0.358920625410974, "entropy": 0.34252429008483887, "sharp": 0.3617107169702649}
+{"time": "16:52:12", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 820.5966508496538, "loss": 0.3727264031767845, "entropy": 0.4652971886098385, "sharp": 0.3102488610893488}
+{"time": "16:55:43", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 822.4657187929031, "loss": 0.5293451454490423, "entropy": 0.39562180638313293, "sharp": 0.35023239254951477}
+{"time": "16:59:17", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 824.7501246198527, "loss": 0.5444122850894928, "entropy": 0.5278748273849487, "sharp": 0.26219596341252327}
+{"time": "17:02:49", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 826.4809989907241, "loss": 0.5787257701158524, "entropy": 0.38239508122205734, "sharp": 0.35344960913062096}
+{"time": "17:06:23", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 827.728820520327, "loss": 0.5460266377776861, "entropy": 0.549994446337223, "sharp": 0.21952258236706257}
+{"time": "17:09:58", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 828.6889326473705, "loss": 0.5219408255070448, "entropy": 0.5463357493281364, "sharp": 0.24708304181694984}
+{"time": "17:13:29", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 829.7989119202435, "loss": 0.3735576132312417, "entropy": 0.4871775917708874, "sharp": 0.27986910194158554}
+{"time": "17:17:05", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 830.7418481888519, "loss": 0.47235487162834033, "entropy": 0.38804525416344404, "sharp": 0.33288342971354723}
+{"time": "17:18:34", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.6529954671859741, "noul_T": 1.56778364750332, "noul_auc": 0.6854399081872036, "noul_acc": 0.642, "noul_f1": 0.6831858407079646, "noul_nll": 0.6503009227886796, "noul_ece": 0.13748677182197572, "noul_brier": 0.22753099334221402, "noul_sharp": 0.18669168651103973, "entropy": 0.5854700803756714}
+{"time": "17:22:11", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 816.2060217077905, "loss": 0.6240489818155766, "entropy": 0.4591265320777893, "sharp": 0.3069760650396347}
+{"time": "17:25:43", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 817.3019600590017, "loss": 0.6629428919404745, "entropy": 0.5767177939414978, "sharp": 0.21505453810095787}
+{"time": "17:29:20", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 818.869037931764, "loss": 0.6016082698479295, "entropy": 0.4876342508941889, "sharp": 0.28366889245808125}
+{"time": "17:32:53", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 820.2346254341536, "loss": 0.4550880156457424, "entropy": 0.46702661365270615, "sharp": 0.30612857080996037}
+{"time": "17:36:23", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 821.1462859269855, "loss": 0.5873807296156883, "entropy": 0.41029493883252144, "sharp": 0.344204718247056}
+{"time": "17:39:53", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 822.530349990488, "loss": 0.5271145382430404, "entropy": 0.4709941688925028, "sharp": 0.2829809635877609}
+{"time": "17:43:31", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 823.360627145859, "loss": 0.6758925393223763, "entropy": 0.5699606537818909, "sharp": 0.20288518257439137}
+{"time": "17:47:04", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 824.0104248296863, "loss": 0.4713939465582371, "entropy": 0.5116083547472954, "sharp": 0.27230210043489933}
+{"time": "17:50:35", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 824.8351839419466, "loss": 0.5855742283165455, "entropy": 0.46853295527398586, "sharp": 0.30818908102810383}
+{"time": "17:54:09", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 825.7169217994069, "loss": 0.4770893193781376, "entropy": 0.46358584985136986, "sharp": 0.2944309590384364}
+{"time": "17:55:37", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.7008872032165527, "noul_T": 2.4896660946033125, "noul_auc": 0.6782501561155087, "noul_acc": 0.528, "noul_f1": 0.4824561403508772, "noul_nll": 0.6934100820273161, "noul_ece": 0.1877158905863762, "noul_brier": 0.2501949381144789, "noul_sharp": 0.11371108889579773, "entropy": 0.6497189402580261}
+{"time": "17:59:01", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 815.7371383378529, "loss": 0.53343066573143, "entropy": 0.5433429330587387, "sharp": 0.2253799457103014}
+{"time": "18:02:31", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 816.3824981195698, "loss": 0.5937585942447186, "entropy": 0.5335094407200813, "sharp": 0.2474512755870819}
+{"time": "18:06:09", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 817.5621641746372, "loss": 0.41736554354429245, "entropy": 0.45776091516017914, "sharp": 0.2996142581105232}
+{"time": "18:09:47", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 818.8375641151256, "loss": 0.5417729634791613, "entropy": 0.44429490715265274, "sharp": 0.3001267910003662}
+{"time": "18:13:08", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 819.2832803332699, "loss": 0.5176213271915913, "entropy": 0.47064588963985443, "sharp": 0.28887568414211273}
+{"time": "18:16:39", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 820.1325920478158, "loss": 0.4833945818245411, "entropy": 0.4966791421175003, "sharp": 0.2836277410387993}
+{"time": "18:20:20", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 821.2181835363507, "loss": 0.5514312759041786, "entropy": 0.5127773433923721, "sharp": 0.26959224976599216}
+{"time": "18:24:00", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 822.1213449988385, "loss": 0.5298340898007154, "entropy": 0.4611290879547596, "sharp": 0.29952465929090977}
+{"time": "18:27:35", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 823.0382058349501, "loss": 0.5084606031887233, "entropy": 0.4768169093877077, "sharp": 0.2965478505939245}
+{"time": "18:31:10", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 823.8042286317592, "loss": 0.3837398439645767, "entropy": 0.43037811666727066, "sharp": 0.3266110494732857}
+{"time": "18:32:38", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.6870254278182983, "noul_T": 1.9037468936859754, "noul_auc": 0.6932034902364518, "noul_acc": 0.564, "noul_f1": 0.5477178423236515, "noul_nll": 0.6801705429777503, "noul_ece": 0.17239673572778705, "noul_brier": 0.24331470918086992, "noul_sharp": 0.14766579866409302, "entropy": 0.6284121870994568}
+{"time": "18:36:08", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 816.5383507638695, "loss": 0.36579227074980736, "entropy": 0.4599296860396862, "sharp": 0.2962314747273922}
+{"time": "18:39:39", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 817.1416104286684, "loss": 0.5723583772778511, "entropy": 0.49667324498295784, "sharp": 0.2748509971424937}
+{"time": "18:43:02", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 817.8289858800566, "loss": 0.7136602364480495, "entropy": 0.5008908687159419, "sharp": 0.279976068995893}
+{"time": "18:46:23", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 818.3574425671809, "loss": 0.49053104408085346, "entropy": 0.5079932045191526, "sharp": 0.2659221328794956}
+{"time": "18:49:59", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 819.2458563512685, "loss": 0.4896436557173729, "entropy": 0.5087234899401665, "sharp": 0.25670865923166275}
+{"time": "18:53:28", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 819.896302069032, "loss": 0.5220542885363102, "entropy": 0.4984864331781864, "sharp": 0.2661774577572942}
+{"time": "18:56:55", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 820.4361379237, "loss": 0.5349264275282621, "entropy": 0.5254762209951878, "sharp": 0.2566579142585397}
+{"time": "19:00:27", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 820.9438069950727, "loss": 0.44816261902451515, "entropy": 0.5012523122131824, "sharp": 0.26441952027380466}
+{"time": "19:04:01", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 821.5717756684161, "loss": 0.5153478682041168, "entropy": 0.5156123712658882, "sharp": 0.2542122099548578}
+{"time": "19:07:37", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 822.0776879186777, "loss": 0.591077296063304, "entropy": 0.4516054503619671, "sharp": 0.29486609995365143}
+{"time": "19:09:05", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.6416885852813721, "noul_T": 1.297941744026611, "noul_auc": 0.6947562066463013, "noul_acc": 0.654, "noul_f1": 0.6916221033868093, "noul_nll": 0.6381101964116096, "noul_ece": 0.11680765664577482, "noul_brier": 0.22335829583073274, "noul_sharp": 0.1659223884344101, "entropy": 0.6084543466567993}
+{"time": "19:12:15", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 8990682, "tok_per_s": 801.8732506453285, "rows_per_s": 1.5021273446902907, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.6236547827720642, "noul_T": 1.2117223117345985, "choice_ce": 0.668276846408844, "choice_T": 1.0711649063804969, "score_ce": 1.0528051853179932, "score_T": 1.0221427041316569, "noul_auc": 0.682496, "noul_acc": 0.648, "noul_f1": 0.7021996615905245, "noul_nll": 0.6334441490061581, "noul_ece": 0.10943237079679967, "noul_brier": 0.22197562435617027, "noul_sharp": 0.1640363186597824, "entropy": 0.6099246740341187}, "dropped_long": 0, "peak_gib": 18.235705852508545}

e16/S1-E19-a2-n4k-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-16 03:12:17

e16/S1-E19-a2-n64k-syn/DONE

@@ -0,0 +1 @@
+2026-10-17 16:49:36

e16/S1-E19-b-n128k-syn/DONE

@@ -0,0 +1 @@
+2026-10-18 18:38:49

e16/S1-E19-b-n16k-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-17 07:56:12

e16/S1-E19-b-n16k-rt-syn/DONE

@@ -0,0 +1 @@
+2026-10-20 04:52:46

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/adapter/README.md

@@ -0,0 +1,206 @@
+---
+base_model: /workspace/jev_kt/models/hf/JEV-9B
+library_name: peft
+tags:
+- base_model:adapter:/workspace/jev_kt/models/hf/JEV-9B
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.21.2

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/adapter/adapter_config.json

@@ -0,0 +1,57 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3_5ForCausalLM",
+    "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
+  },
+  "base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "kasa_config": null,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.21.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "q_proj",
+    "in_proj_qkv",
+    "in_proj_z",
+    "out_proj",
+    "k_proj",
+    "v_proj",
+    "up_proj",
+    "o_proj",
+    "down_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": null,
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/adapter/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf36bd762bba75ac7d28d18898576d0981294b11ccb1f507692eb701787e5885
+size 160486456

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/adapter_vllm/adapter_config.json

@@ -0,0 +1,58 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3_5ForCausalLM",
+    "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
+  },
+  "base_model_name_or_path": null,
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "kasa_config": null,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64.0,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.21.0",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "in_proj_qkv",
+    "in_proj_z",
+    "k_proj",
+    "lm_head",
+    "o_proj",
+    "out_proj",
+    "q_proj",
+    "up_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": null,
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf500137bba7449ec595367ab081a552921ff21a838ed074ca85679864354cbe
+size 176641832

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/adapter_vllm/decision_head.json

@@ -0,0 +1,99 @@
+{
+ "bias": [
+  0.00041853307629935443,
+  -0.0004187873564660549,
+  -0.002389779780060053,
+  -0.0019469341496005654,
+  -0.0005223387270234525,
+  0.0019261763663962483,
+  0.0013853152049705386,
+  -0.0023466842249035835,
+  7.24989440641366e-05,
+  -0.00011821574298664927,
+  3.653583553386852e-05,
+  0.0009474153630435467,
+  -0.0013647503219544888,
+  -0.0034368657507002354,
+  -0.0040892669931054115,
+  -0.005190850701183081,
+  -0.005747793707996607,
+  -0.004366291221231222,
+  -0.004934487398713827,
+  -0.004979669116437435,
+  -0.0037483018822968006,
+  -0.003373329062014818,
+  -0.0057093421928584576,
+  -0.0034768348559737206
+ ],
+ "verbalizer_ids": [
+  3721,
+  1802,
+  15,
+  16,
+  17,
+  18,
+  19,
+  20,
+  32,
+  33,
+  34,
+  35,
+  36,
+  37,
+  38,
+  39,
+  40,
+  41,
+  42,
+  43,
+  44,
+  45,
+  46,
+  47
+ ],
+ "slots": {
+  "num_slots": 24,
+  "ranges": {
+   "noul": [
+    0,
+    2
+   ],
+   "score": [
+    2,
+    8
+   ],
+   "choice": [
+    8,
+    24
+   ]
+  },
+  "verbalizers": [
+   "false",
+   "true",
+   "0",
+   "1",
+   "2",
+   "3",
+   "4",
+   "5",
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "template_version": "bare-v1"
+ },
+ "note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
+}

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/calibration.json

@@ -0,0 +1,29 @@
+{
+ "version": 1,
+ "per_kind": {
+  "noul": 0.8135197452442088,
+  "choice": 1.06025618459646,
+  "score": 0.9688074232706271
+ },
+ "fit": {
+  "noul": {
+   "n": 1087,
+   "T": 0.8135197452442088,
+   "ce_before": 0.5723596215248108,
+   "ce_after": 0.5692269802093506
+  },
+  "choice": {
+   "n": 60,
+   "T": 1.06025618459646,
+   "ce_before": 0.6637058854103088,
+   "ce_after": 0.6629296541213989
+  },
+  "score": {
+   "n": 53,
+   "T": 0.9688074232706271,
+   "ce_before": 1.051501750946045,
+   "ce_after": 1.0511925220489502
+  }
+ },
+ "source": "models/e16/S1-E19-b-n16k-rt-syn/ckpt-16000"
+}

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/head.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe97b783c9f97a608f6b996490c06e29c9cf7e847798038c15cace7535460316
+size 393464

e16/S1-E19-b-n16k-rt-syn/ckpt-16000/judge_config.json

@@ -0,0 +1,164 @@
+{
+ "base_model_path": "/root/models/Qwen3.5-9B",
+ "hidden_size": 4096,
+ "slots": {
+  "num_slots": 24,
+  "ranges": {
+   "noul": [
+    0,
+    2
+   ],
+   "score": [
+    2,
+    8
+   ],
+   "choice": [
+    8,
+    24
+   ]
+  },
+  "verbalizers": [
+   "false",
+   "true",
+   "0",
+   "1",
+   "2",
+   "3",
+   "4",
+   "5",
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "template_version": "bare-v1"
+ },
+ "verbalizer_ids": [
+  3721,
+  1802,
+  15,
+  16,
+  17,
+  18,
+  19,
+  20,
+  32,
+  33,
+  34,
+  35,
+  36,
+  37,
+  38,
+  39,
+  40,
+  41,
+  42,
+  43,
+  44,
+  45,
+  46,
+  47
+ ],
+ "kinds": [
+  "noul",
+  "choice",
+  "score"
+ ],
+ "model_name": "jev-judge-qwen35-9b",
+ "model_version": "0.8.0",
+ "lm_head_included": true,
+ "weights_mode": "unmerged",
+ "adapter_subfolder": "adapter",
+ "lora": {
+  "r": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "target_modules": [
+   "down_proj",
+   "gate_proj",
+   "in_proj_qkv",
+   "in_proj_z",
+   "k_proj",
+   "o_proj",
+   "out_proj",
+   "q_proj",
+   "up_proj",
+   "v_proj"
+  ]
+ },
+ "trained_stage": "e16",
+ "source_checkpoint": "models/e16/S1-E19-b-n16k-rt-syn/ckpt-16000",
+ "softcap": null,
+ "e16": {
+  "data": "data/e19/E19-b-n16k",
+  "data_manifest": {
+   "arm": "E19-b-n16k",
+   "name": null,
+   "train_sha256": "9e31412d9aac0064859cdf9aae1636242d1a9dbfc3ab398a1f5ec10da1e7a50e",
+   "n_train": 16000
+  },
+  "init": "released",
+  "objective": "s-log",
+  "sigma": 0.5,
+  "K": 8,
+  "reward": "brier",
+  "clip": 0.2,
+  "passes": 1,
+  "lr": 0.0001,
+  "lr_head": 0.0002,
+  "rows_per_step": 32,
+  "train_seed": 0,
+  "replay": "data/e16/replay/train.jsonl",
+  "replay_frac": 0.05,
+  "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
+  "max_len": 3072,
+  "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps",
+  "total_rows": 16000,
+  "lora": {
+   "r": 16,
+   "lora_alpha": 32,
+   "lora_dropout": 0.05
+  },
+  "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
+  "event": "ckpt",
+  "tag": "ckpt-16000",
+  "rows_seen": 16000,
+  "replay_seen": 842,
+  "step": 500,
+  "tokens": 9808867,
+  "tok_per_s": 812.4848276388393,
+  "rows_per_s": 1.395050973921692,
+  "val": {
+   "n": 1200,
+   "dropped_long": 0,
+   "noul_ce": 0.5723596215248108,
+   "noul_T": 0.8135197452442088,
+   "choice_ce": 0.6637058854103088,
+   "choice_T": 1.06025618459646,
+   "score_ce": 1.051501750946045,
+   "score_T": 0.9688074232706271,
+   "noul_auc": 0.7213738666666666,
+   "noul_acc": 0.69,
+   "noul_f1": 0.7707100591715976,
+   "noul_nll": 0.5888399675711989,
+   "noul_ece": 0.04534497305750847,
+   "noul_brier": 0.2018144641243176,
+   "noul_sharp": 0.16441088914871216,
+   "entropy": 0.6122345924377441
+  },
+  "dropped_long": 0,
+  "peak_gib": 18.236003875732422
+ }
+}

e16/S1-E19-b-n16k-rt-syn/train_log.jsonl

@@ -0,0 +1,57 @@
+{"time": "01:31:31", "event": "start", "data": "data/e19/E19-b-n16k", "data_manifest": {"arm": "E19-b-n16k", "name": null, "train_sha256": "9e31412d9aac0064859cdf9aae1636242d1a9dbfc3ab398a1f5ec10da1e7a50e", "n_train": 16000}, "init": "released", "objective": "s-log", "sigma": 0.5, "K": 8, "reward": "brier", "clip": 0.2, "passes": 1, "lr": 0.0001, "lr_head": 0.0002, "rows_per_step": 32, "train_seed": 0, "replay": "data/e16/replay/train.jsonl", "replay_frac": 0.05, "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e", "max_len": 3072, "schedule": "warmup 0.03 + cosine to x0.05 over 500 steps", "total_rows": 16000, "lora": {"r": 16, "lora_alpha": 32, "lora_dropout": 0.05}, "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4", "device": "cuda", "load_s": 79.34462976455688}
+{"time": "01:35:22", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 840.7661624934583, "loss": 0.6416471749544144, "entropy": 0.6679874174296856, "sharp": 0.09398072306066751}
+{"time": "01:38:58", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 846.490472932749, "loss": 0.6551759392023087, "entropy": 0.6430379413068295, "sharp": 0.12173190526664257}
+{"time": "01:42:54", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 851.4914282719631, "loss": 0.7106168419122696, "entropy": 0.6584099866449833, "sharp": 0.11426412966102362}
+{"time": "01:46:39", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 853.3469278820797, "loss": 0.6800252795219421, "entropy": 0.6758302748203278, "sharp": 0.08097615092992783}
+{"time": "01:50:26", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 854.4702631370465, "loss": 0.6574342548847198, "entropy": 0.6583614125847816, "sharp": 0.1017528185620904}
+{"time": "01:54:14", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 856.1254685596642, "loss": 0.6474643759429455, "entropy": 0.6452968418598175, "sharp": 0.12556627485901117}
+{"time": "01:58:04", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 856.6372583207745, "loss": 0.6487826257944107, "entropy": 0.6831155568361282, "sharp": 0.057710735127329826}
+{"time": "02:01:54", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 856.5255069935561, "loss": 0.7086809352040291, "entropy": 0.6393271498382092, "sharp": 0.13126030378043652}
+{"time": "02:05:40", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 856.998930195703, "loss": 0.6685285903513432, "entropy": 0.6669215448200703, "sharp": 0.10060268267989159}
+{"time": "02:09:27", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 856.252187442035, "loss": 0.6253704205155373, "entropy": 0.661177471280098, "sharp": 0.10144310351461172}
+{"time": "02:10:55", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5964431762695312, "noul_T": 0.8831745654077745, "noul_auc": 0.7117010683363993, "noul_acc": 0.688, "noul_f1": 0.75625, "noul_nll": 0.6003437442332507, "noul_ece": 0.03960577178001404, "noul_brier": 0.2066063846506394, "noul_sharp": 0.16483941674232483, "entropy": 0.6129420399665833}
+{"time": "02:14:48", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 827.9876186723793, "loss": 0.6816069111227989, "entropy": 0.6382579281926155, "sharp": 0.13827434927225113}
+{"time": "02:18:28", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 829.8853486215241, "loss": 0.6276846416294575, "entropy": 0.6610538326203823, "sharp": 0.10919279884546995}
+{"time": "02:22:19", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 832.1608961933646, "loss": 0.6449321880936623, "entropy": 0.671237125992775, "sharp": 0.08378801215440035}
+{"time": "02:26:14", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 833.5243989504289, "loss": 0.6575691029429436, "entropy": 0.65861926227808, "sharp": 0.11133739259094}
+{"time": "02:29:58", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 835.0050475252322, "loss": 0.6752120107412338, "entropy": 0.6664692014455795, "sharp": 0.08565635699778795}
+{"time": "02:33:50", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 836.5276400297607, "loss": 0.7220019102096558, "entropy": 0.6175351366400719, "sharp": 0.16432397067546844}
+{"time": "02:37:45", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 837.8800578342609, "loss": 0.6397408656775951, "entropy": 0.6512692756950855, "sharp": 0.12000797595828772}
+{"time": "02:41:36", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 838.7248322390666, "loss": 0.6728365942835808, "entropy": 0.6538522988557816, "sharp": 0.11869410052895546}
+{"time": "02:45:22", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 839.244121752934, "loss": 0.64999870210886, "entropy": 0.654944472014904, "sharp": 0.10712145920842886}
+{"time": "02:49:13", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 840.1310495772577, "loss": 0.6845478937029839, "entropy": 0.6774301528930664, "sharp": 0.06942567136138678}
+{"time": "02:50:41", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.6271136403083801, "noul_T": 0.586638778266236, "noul_auc": 0.7377597002582235, "noul_acc": 0.662, "noul_f1": 0.6943942133815552, "noul_nll": 0.6235430086255074, "noul_ece": 0.12544882184267042, "noul_brier": 0.2165295909018273, "noul_sharp": 0.10261821001768112, "entropy": 0.6627551913261414}
+{"time": "02:54:30", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 825.8320986869265, "loss": 0.6748520731925964, "entropy": 0.6297518685460091, "sharp": 0.15004642121493816}
+{"time": "02:58:15", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 827.378174661144, "loss": 0.6887126602232456, "entropy": 0.6604589074850082, "sharp": 0.10578704159706831}
+{"time": "03:02:03", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 828.6369955420482, "loss": 0.6807634234428406, "entropy": 0.6513927578926086, "sharp": 0.12257815152406693}
+{"time": "03:05:57", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 829.7534272327645, "loss": 0.6757249608635902, "entropy": 0.6517044082283974, "sharp": 0.11861236020922661}
+{"time": "03:09:45", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 830.6757755296687, "loss": 0.7071560956537724, "entropy": 0.6361428834497929, "sharp": 0.14567251317203045}
+{"time": "03:13:33", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 831.3294961215616, "loss": 0.6789720952510834, "entropy": 0.6721667125821114, "sharp": 0.08510761987417936}
+{"time": "03:17:32", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 832.5692936584524, "loss": 0.6502889152616262, "entropy": 0.6423392742872238, "sharp": 0.13550199195742607}
+{"time": "03:21:22", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 833.3837441091671, "loss": 0.6975204311311245, "entropy": 0.6784060820937157, "sharp": 0.07007143460214138}
+{"time": "03:25:16", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 834.2972440408065, "loss": 0.666705098003149, "entropy": 0.6606337241828442, "sharp": 0.10764894261956215}
+{"time": "03:29:09", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 835.0841686706652, "loss": 0.674651738256216, "entropy": 0.6482210867106915, "sharp": 0.1335096014663577}
+{"time": "03:30:37", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5930076241493225, "noul_T": 0.6900405565484261, "noul_auc": 0.7367301817690841, "noul_acc": 0.71, "noul_f1": 0.7758887171561051, "noul_nll": 0.5907440386712551, "noul_ece": 0.061869306087493914, "noul_brier": 0.2013389410117992, "noul_sharp": 0.14554060995578766, "entropy": 0.6348153948783875}
+{"time": "03:34:23", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 825.676836655453, "loss": 0.5999349504709244, "entropy": 0.665407981723547, "sharp": 0.10429427959024906}
+{"time": "03:38:11", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 826.5217586371051, "loss": 0.6684055104851723, "entropy": 0.6669856086373329, "sharp": 0.09141119290143251}
+{"time": "03:42:06", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 827.4647143810705, "loss": 0.695707505568862, "entropy": 0.6447131410241127, "sharp": 0.12685945630073547}
+{"time": "03:45:49", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 828.1200362828814, "loss": 0.6759919933974743, "entropy": 0.6487890593707561, "sharp": 0.11696403473615646}
+{"time": "03:49:39", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 828.9706371176582, "loss": 0.6452272497117519, "entropy": 0.655936699360609, "sharp": 0.11084896977990866}
+{"time": "03:53:26", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 829.5822524408109, "loss": 0.6544655114412308, "entropy": 0.6363691687583923, "sharp": 0.1469872109591961}
+{"time": "03:57:21", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 830.3068580055784, "loss": 0.6865418404340744, "entropy": 0.638183668255806, "sharp": 0.13787089474499226}
+{"time": "04:01:09", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 830.9034194194427, "loss": 0.6698485910892487, "entropy": 0.6436968371272087, "sharp": 0.12814036756753922}
+{"time": "04:05:02", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 831.8728006378307, "loss": 0.6417997851967812, "entropy": 0.6429427266120911, "sharp": 0.1223332742229104}
+{"time": "04:08:41", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 832.4552561691451, "loss": 0.6876776963472366, "entropy": 0.6284196823835373, "sharp": 0.14502350706607103}
+{"time": "04:10:09", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5873317122459412, "noul_T": 0.8172684020234544, "noul_auc": 0.7343842298020287, "noul_acc": 0.71, "noul_f1": 0.7730829420970265, "noul_nll": 0.5875646459609271, "noul_ece": 0.049862171888351445, "noul_brier": 0.20041123678481845, "noul_sharp": 0.16171549260616302, "entropy": 0.6141359210014343}
+{"time": "04:13:56", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 825.4999736297234, "loss": 0.6560182794928551, "entropy": 0.6658743768930435, "sharp": 0.09087194968014956}
+{"time": "04:17:40", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 826.3680545344017, "loss": 0.6620576977729797, "entropy": 0.6541808471083641, "sharp": 0.11118748597800732}
+{"time": "04:21:33", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 827.1773821802286, "loss": 0.6665778607130051, "entropy": 0.6465357467532158, "sharp": 0.1227754270657897}
+{"time": "04:25:21", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 827.8440116905582, "loss": 0.6667315606027842, "entropy": 0.6536180730909109, "sharp": 0.11583110596984625}
+{"time": "04:29:10", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 828.6482165661197, "loss": 0.645016323775053, "entropy": 0.6607457809150219, "sharp": 0.0980006130412221}
+{"time": "04:33:00", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 829.2298580494971, "loss": 0.6537435129284859, "entropy": 0.652746357023716, "sharp": 0.11425115168094635}
+{"time": "04:36:55", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 829.8939265476491, "loss": 0.720472251996398, "entropy": 0.6619589664041996, "sharp": 0.10576295014470816}
+{"time": "04:40:41", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 830.4927911041059, "loss": 0.6888858638703823, "entropy": 0.6586914137005806, "sharp": 0.10336506646126509}
+{"time": "04:44:23", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 830.9565615541648, "loss": 0.5681696087121964, "entropy": 0.6408889293670654, "sharp": 0.13228241726756096}
+{"time": "04:48:08", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 831.4687967613543, "loss": 0.6161541864275932, "entropy": 0.6481572985649109, "sharp": 0.1234055906534195}
+{"time": "04:49:36", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5878974795341492, "noul_T": 0.7946635749095625, "noul_auc": 0.7321226645963781, "noul_acc": 0.7, "noul_f1": 0.7692307692307693, "noul_nll": 0.5894184568673372, "noul_ece": 0.06553434431552889, "noul_brier": 0.20128141626002066, "noul_sharp": 0.1585736721754074, "entropy": 0.6175200343132019}
+{"time": "04:52:46", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9808867, "tok_per_s": 812.4848276388393, "rows_per_s": 1.395050973921692, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5723596215248108, "noul_T": 0.8135197452442088, "choice_ce": 0.6637058854103088, "choice_T": 1.06025618459646, "score_ce": 1.051501750946045, "score_T": 0.9688074232706271, "noul_auc": 0.7213738666666666, "noul_acc": 0.69, "noul_f1": 0.7707100591715976, "noul_nll": 0.5888399675711989, "noul_ece": 0.04534497305750847, "noul_brier": 0.2018144641243176, "noul_sharp": 0.16441088914871216, "entropy": 0.6122345924377441}, "dropped_long": 0, "peak_gib": 18.236003875732422}

e16/S1-E19-b-n16k-soft-a100-syn/DONE

@@ -0,0 +1 @@
+2026-10-16 09:12:55

e16/S1-E19-b-n256k-syn/DONE

@@ -0,0 +1 @@
+2026-10-19 19:31:44

e16/S1-E19-b-n256k-syn/ckpt-256000/adapter/README.md

@@ -0,0 +1,206 @@
+---
+base_model: /workspace/jev_kt/models/hf/JEV-9B
+library_name: peft
+tags:
+- base_model:adapter:/workspace/jev_kt/models/hf/JEV-9B
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.21.2

e16/S1-E19-b-n256k-syn/ckpt-256000/adapter/adapter_config.json

@@ -0,0 +1,57 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3_5ForCausalLM",
+    "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
+  },
+  "base_model_name_or_path": "/workspace/jev_kt/models/hf/JEV-9B",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "kasa_config": null,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.21.0",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "in_proj_z",
+    "up_proj",
+    "in_proj_qkv",
+    "k_proj",
+    "out_proj",
+    "o_proj",
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": null,
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

e16/S1-E19-b-n256k-syn/ckpt-256000/adapter/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6ff4f018e50cb9dafdda5bab731a3ea9744b564342d6927b9539b0977f8b4a7
+size 160486456

e16/S1-E19-b-n256k-syn/ckpt-256000/adapter_vllm/adapter_config.json

@@ -0,0 +1,58 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "Qwen3_5ForCausalLM",
+    "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
+  },
+  "base_model_name_or_path": null,
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "kasa_config": null,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64.0,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "monteclora_config": null,
+  "peft_type": "LORA",
+  "peft_version": "0.21.0",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "down_proj",
+    "gate_proj",
+    "in_proj_qkv",
+    "in_proj_z",
+    "k_proj",
+    "lm_head",
+    "o_proj",
+    "out_proj",
+    "q_proj",
+    "up_proj",
+    "v_proj"
+  ],
+  "target_parameters": null,
+  "task_type": null,
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false,
+  "velora_config": null
+}

e16/S1-E19-b-n256k-syn/ckpt-256000/adapter_vllm/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2443af36df98fcb432de000780d31735ce1737b2d2a971750f4b0fd0e25b3b34
+size 176641832

e16/S1-E19-b-n256k-syn/ckpt-256000/adapter_vllm/decision_head.json

@@ -0,0 +1,99 @@
+{
+ "bias": [
+  0.0008895411738194525,
+  -0.0008899205131456256,
+  -0.006735881324857473,
+  -0.005502605810761452,
+  -0.0029139602556824684,
+  0.0021956299897283316,
+  0.0013801128370687366,
+  -0.005935822147876024,
+  -0.002272534416988492,
+  -0.0017964727012440562,
+  -0.002017927821725607,
+  -0.0024679864291101694,
+  -0.006912675686180592,
+  -0.00539945624768734,
+  -0.011302140541374683,
+  -0.008620599284768105,
+  -0.008465982973575592,
+  -0.009283804334700108,
+  -0.009762704372406006,
+  -0.008379790000617504,
+  -0.00788922794163227,
+  -0.008381989784538746,
+  -0.010746600106358528,
+  -0.008571792393922806
+ ],
+ "verbalizer_ids": [
+  3721,
+  1802,
+  15,
+  16,
+  17,
+  18,
+  19,
+  20,
+  32,
+  33,
+  34,
+  35,
+  36,
+  37,
+  38,
+  39,
+  40,
+  41,
+  42,
+  43,
+  44,
+  45,
+  46,
+  47
+ ],
+ "slots": {
+  "num_slots": 24,
+  "ranges": {
+   "noul": [
+    0,
+    2
+   ],
+   "score": [
+    2,
+    8
+   ],
+   "choice": [
+    8,
+    24
+   ]
+  },
+  "verbalizers": [
+   "false",
+   "true",
+   "0",
+   "1",
+   "2",
+   "3",
+   "4",
+   "5",
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "template_version": "bare-v1"
+ },
+ "note": "decision logits = lm_head-LoRA logprobs of verbalizer ids + bias, then per-kind temperature"
+}

e16/S1-E19-b-n256k-syn/ckpt-256000/calibration.json

@@ -0,0 +1,29 @@
+{
+ "version": 1,
+ "per_kind": {
+  "noul": 0.7219452547802345,
+  "choice": 0.9974698100810453,
+  "score": 0.9681490341363794
+ },
+ "fit": {
+  "noul": {
+   "n": 1087,
+   "T": 0.7219452547802345,
+   "ce_before": 0.5737836360931396,
+   "ce_after": 0.5665724873542786
+  },
+  "choice": {
+   "n": 60,
+   "T": 0.9974698100810453,
+   "ce_before": 0.6358097791671753,
+   "ce_after": 0.6358082890510559
+  },
+  "score": {
+   "n": 53,
+   "T": 0.9681490341363794,
+   "ce_before": 1.0383120775222778,
+   "ce_after": 1.0379929542541504
+  }
+ },
+ "source": "models/e16/S1-E19-b-n256k-syn/ckpt-256000"
+}

e16/S1-E19-b-n256k-syn/ckpt-256000/head.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aaf78ec14ccc859f253f5ad69defbe924270592a97a5dbfdd952416606dcd4fc
+size 393464

e16/S1-E19-b-n256k-syn/ckpt-256000/judge_config.json

@@ -0,0 +1,164 @@
+{
+ "base_model_path": "/root/models/Qwen3.5-9B",
+ "hidden_size": 4096,
+ "slots": {
+  "num_slots": 24,
+  "ranges": {
+   "noul": [
+    0,
+    2
+   ],
+   "score": [
+    2,
+    8
+   ],
+   "choice": [
+    8,
+    24
+   ]
+  },
+  "verbalizers": [
+   "false",
+   "true",
+   "0",
+   "1",
+   "2",
+   "3",
+   "4",
+   "5",
+   "A",
+   "B",
+   "C",
+   "D",
+   "E",
+   "F",
+   "G",
+   "H",
+   "I",
+   "J",
+   "K",
+   "L",
+   "M",
+   "N",
+   "O",
+   "P"
+  ],
+  "template_version": "bare-v1"
+ },
+ "verbalizer_ids": [
+  3721,
+  1802,
+  15,
+  16,
+  17,
+  18,
+  19,
+  20,
+  32,
+  33,
+  34,
+  35,
+  36,
+  37,
+  38,
+  39,
+  40,
+  41,
+  42,
+  43,
+  44,
+  45,
+  46,
+  47
+ ],
+ "kinds": [
+  "noul",
+  "choice",
+  "score"
+ ],
+ "model_name": "jev-judge-qwen35-9b",
+ "model_version": "0.8.0",
+ "lm_head_included": true,
+ "weights_mode": "unmerged",
+ "adapter_subfolder": "adapter",
+ "lora": {
+  "r": 16,
+  "lora_alpha": 32,
+  "lora_dropout": 0.05,
+  "target_modules": [
+   "down_proj",
+   "gate_proj",
+   "in_proj_qkv",
+   "in_proj_z",
+   "k_proj",
+   "o_proj",
+   "out_proj",
+   "q_proj",
+   "up_proj",
+   "v_proj"
+  ]
+ },
+ "trained_stage": "e16",
+ "source_checkpoint": "models/e16/S1-E19-b-n256k-syn/ckpt-256000",
+ "softcap": null,
+ "e16": {
+  "data": "data/e19/E19-b-n256k",
+  "data_manifest": {
+   "arm": "E19-b-n256k",
+   "name": null,
+   "train_sha256": "0c9b39bf9b638138898c8b150dd81bf7e3d1e30f1088b4b466dac1a96eea0519",
+   "n_train": 256000
+  },
+  "init": "released",
+  "objective": "s-log",
+  "sigma": 0.5,
+  "K": 8,
+  "reward": "brier",
+  "clip": 0.2,
+  "passes": 1,
+  "lr": 0.0001,
+  "lr_head": 0.0002,
+  "rows_per_step": 32,
+  "train_seed": 0,
+  "replay": "data/e16/replay/train.jsonl",
+  "replay_frac": 0.05,
+  "replay_sha256": "9dd2df19851678750cf79d2fdb47581720930ef2f0c6e18e6fbe23ef2181847e",
+  "max_len": 3072,
+  "schedule": "warmup 0.03 + cosine to x0.05 over 8000 steps",
+  "total_rows": 256000,
+  "lora": {
+   "r": 16,
+   "lora_alpha": 32,
+   "lora_dropout": 0.05
+  },
+  "trainer_sha256": "bc7c08e836bd99bb2665b02903cd41404ea029ecfa9b68a9ec1a7cda96958ba4",
+  "event": "ckpt",
+  "tag": "ckpt-256000",
+  "rows_seen": 256000,
+  "replay_seen": 13473,
+  "step": 8000,
+  "tokens": 157430651,
+  "tok_per_s": 865.2182058016194,
+  "rows_per_s": 1.4809882579083644,
+  "val": {
+   "n": 1200,
+   "dropped_long": 0,
+   "noul_ce": 0.5737836360931396,
+   "noul_T": 0.7219452547802345,
+   "choice_ce": 0.6358097791671753,
+   "choice_T": 0.9974698100810453,
+   "score_ce": 1.0383120775222778,
+   "score_T": 0.9681490341363794,
+   "noul_auc": 0.7238229333333334,
+   "noul_acc": 0.685,
+   "noul_f1": 0.7629796839729119,
+   "noul_nll": 0.5908580548092723,
+   "noul_ece": 0.05055843117833138,
+   "noul_brier": 0.20239246989918638,
+   "noul_sharp": 0.15035562217235565,
+   "entropy": 0.6265034675598145
+  },
+  "dropped_long": 0,
+  "peak_gib": 18.23600435256958
+ }
+}