diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/DONE b/e41/S1-E41-FRESH-A050-ts0-syn/DONE new file mode 100644 index 0000000000000000000000000000000000000000..829ff0ad27d1b35484f9af91f347a1f66044f33a --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/DONE @@ -0,0 +1 @@ +2026-11-01 09:27:42 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/README.md b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/README.md new file mode 100644 index 0000000000000000000000000000000000000000..57167079ef3f181aacdbd678a733d71697d18f63 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-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 + + + + + +## Model Details + +### Model Description + + + + + +- **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] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +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 + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +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] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[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 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/adapter_config.json b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..2ebc4c5323131cec3365c0bceb4f9682bad5fae9 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-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": "0.21.2", + "qalora_group_size": 16, + "r": 16, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "o_proj", + "gate_proj", + "k_proj", + "in_proj_z", + "in_proj_qkv", + "up_proj", + "down_proj", + "v_proj", + "q_proj", + "out_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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..78312e8f8e09ad4b356d014e4e4e6adeb4c269c7 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7158c69f9016947b6f26c0381f5b821d892ae224509a566e223743c08935ffa +size 160486456 diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/adapter_config.json b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..fadf1add9c5135762b687d4f5260deb938741ff5 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-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": "0.21.2", + "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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..686418ed69e06c2cc8c592ae7c8c8508a476f97b --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2b24af2d87696471f1b89094658877c5dde71edd905bef9597140c0fec37224 +size 176641832 diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json new file mode 100644 index 0000000000000000000000000000000000000000..6db462e749551b26798123ab1f583b9e0222a276 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json @@ -0,0 +1,99 @@ +{ + "bias": [ + -6.5171880123671144e-06, + 6.5036815612984356e-06, + -0.0003035451809410006, + -0.0006027663475833833, + -0.00019296040409244597, + 0.0006487983046099544, + 0.000769275997299701, + -0.0010308213531970978, + -0.0002992550143972039, + -0.0002714841393753886, + 0.0001120415108744055, + 1.3407094229478389e-05, + -0.0005220042075961828, + -0.0008380978251807392, + -0.0036718491464853287, + -0.003459431231021881, + -0.0005925728473812342, + -0.0030832898337394, + -0.0030681416392326355, + -0.003155431477352977, + -0.002447268459945917, + -0.002963721053674817, + -0.003133098129183054, + -0.0030540970619767904 + ], + "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" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/calibration.json b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/calibration.json new file mode 100644 index 0000000000000000000000000000000000000000..fc36e18ebcb742a829cae3330c1d276c18dac18b --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/calibration.json @@ -0,0 +1,29 @@ +{ + "version": 1, + "per_kind": { + "noul": 0.8311502624434424, + "choice": 0.9575453686506029, + "score": 0.9968860395022641 + }, + "fit": { + "noul": { + "n": 1087, + "T": 0.8311502624434424, + "ce_before": 0.5641881227493286, + "ce_after": 0.5613892674446106 + }, + "choice": { + "n": 60, + "T": 0.9575453686506029, + "ce_before": 0.9396607279777527, + "ce_after": 0.9392049312591553 + }, + "score": { + "n": 53, + "T": 0.9968860395022641, + "ce_before": 1.273746371269226, + "ce_after": 1.27374267578125 + } + }, + "source": "models/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/head.safetensors b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/head.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b2aa778c435cff33ffc995805d0d62d9b87e55c2 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/head.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c5264bd1a012f8939fb0ab29fd3e62a367d29af0cc51c4d4561ab0f71f6a994 +size 393464 diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/judge_config.json b/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000/judge_config.json new file mode 100644 index 0000000000000000000000000000000000000000..cec3603302558e414503b778c55c9b117e4f5305 --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-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/e41/S1-E41-FRESH-A050-ts0-syn/ckpt-16000", + "softcap": null, + "e16": { + "data": "data/e18/G-d512-n16k", + "data_manifest": { + "arm": "G-d512-n16k", + "name": null, + "train_sha256": "ef94315960694ce58bdc6d61b8665b7bf46f4e2996b9e82503e1d17bb87a14b7", + "n_train": 16000 + }, + "init": "fresh", + "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", + "event": "ckpt", + "tag": "ckpt-16000", + "rows_seen": 16000, + "replay_seen": 842, + "step": 500, + "tokens": 9113372, + "tok_per_s": 828.5527392816117, + "rows_per_s": 1.5312098779377614, + "val": { + "n": 1200, + "dropped_long": 0, + "noul_ce": 0.5641881227493286, + "noul_T": 0.8311502624434424, + "choice_ce": 0.9396607279777527, + "choice_T": 0.9575453686506029, + "score_ce": 1.273746371269226, + "score_T": 0.9968860395022641, + "noul_auc": 0.7455445333333334, + "noul_acc": 0.698, + "noul_f1": 0.770516717325228, + "noul_nll": 0.5714521871805192, + "noul_ece": 0.03739325839281083, + "noul_brier": 0.19463277056479883, + "noul_sharp": 0.17946934700012207, + "entropy": 0.5962491631507874 + }, + "dropped_long": 0, + "peak_gib": 18.237416744232178 + } +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A050-ts0-syn/train_log.jsonl b/e41/S1-E41-FRESH-A050-ts0-syn/train_log.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..21619cd9c879b2e4893200443d66894c0555511c --- /dev/null +++ b/e41/S1-E41-FRESH-A050-ts0-syn/train_log.jsonl @@ -0,0 +1,57 @@ +{"time": "06:24:19", "event": "start", "data": "data/e18/G-d512-n16k", "data_manifest": {"arm": "G-d512-n16k", "name": null, "train_sha256": "ef94315960694ce58bdc6d61b8665b7bf46f4e2996b9e82503e1d17bb87a14b7", "n_train": 16000}, "init": "fresh", "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", "device": "cuda", "load_s": 84.93021035194397} +{"time": "06:27:48", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 862.7680979551621, "loss": 0.6780711822211742, "entropy": 0.4750773645937443, "sharp": 0.29822914861142635} +{"time": "06:31:17", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 865.9284350867372, "loss": 0.6622143499553204, "entropy": 0.6432023905217648, "sharp": 0.12267992086708546} +{"time": "06:34:48", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 868.1183346302579, "loss": 0.6988406348973513, "entropy": 0.4990328270941973, "sharp": 0.2543744817376137} +{"time": "06:38:12", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 869.8211898437319, "loss": 0.7187316119670868, "entropy": 0.6215901337563992, "sharp": 0.16205210611224174} +{"time": "06:41:44", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 871.2518618594983, "loss": 0.6524413619190454, "entropy": 0.6527555659413338, "sharp": 0.12186353281140327} +{"time": "06:45:08", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 873.2330290237377, "loss": 0.6079571135342121, "entropy": 0.5565574392676353, "sharp": 0.22341892682015896} +{"time": "06:48:38", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 873.1316650661036, "loss": 0.6517663598060608, "entropy": 0.6045583337545395, "sharp": 0.17981518991291523} +{"time": "06:52:06", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 873.2250137384335, "loss": 0.6245818436145782, "entropy": 0.6153163127601147, "sharp": 0.17324366234242916} +{"time": "06:55:39", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 874.5600879353797, "loss": 0.7294809464365244, "entropy": 0.5931927897036076, "sharp": 0.18030779715627432} +{"time": "06:59:09", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 874.1231067864304, "loss": 0.5808363053947687, "entropy": 0.6224967427551746, "sharp": 0.1607719026505947} +{"time": "07:00:33", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5740360021591187, "noul_T": 0.9558321510733195, "noul_auc": 0.7515147423672174, "noul_acc": 0.712, "noul_f1": 0.7837837837837838, "noul_nll": 0.5728424141779542, "noul_ece": 0.06566110587120057, "noul_brier": 0.1949549614731103, "noul_sharp": 0.19508957862854004, "entropy": 0.5818634629249573} +{"time": "07:03:58", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 844.3461118854674, "loss": 0.6429532207548618, "entropy": 0.5802754387259483, "sharp": 0.20042063668370247} +{"time": "07:07:36", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 847.8860014997839, "loss": 0.5994194597005844, "entropy": 0.5494805630296469, "sharp": 0.22450928390026093} +{"time": "07:11:13", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 850.8085967059347, "loss": 0.5950643680989742, "entropy": 0.5561755187809467, "sharp": 0.22023704834282398} +{"time": "07:14:43", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 852.3060875294714, "loss": 0.6320975795388222, "entropy": 0.6572491899132729, "sharp": 0.1121342247352004} +{"time": "07:18:15", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 853.8255768079597, "loss": 0.6728186272084713, "entropy": 0.6449480690062046, "sharp": 0.1278781844303012} +{"time": "07:21:38", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 855.2382803246161, "loss": 0.6514099668711424, "entropy": 0.5722448937594891, "sharp": 0.20415380038321018} +{"time": "07:25:10", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 856.4760476806522, "loss": 0.6613807119429111, "entropy": 0.6127980314195156, "sharp": 0.16541082318872213} +{"time": "07:28:38", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 857.7935799223224, "loss": 0.6313244663178921, "entropy": 0.6178101301193237, "sharp": 0.16268466413021088} +{"time": "07:32:06", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 858.5175201250418, "loss": 0.664512861520052, "entropy": 0.6208578757941723, "sharp": 0.15545270405709743} +{"time": "07:35:36", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 859.4465307862027, "loss": 0.5887272749096155, "entropy": 0.5815163925290108, "sharp": 0.20409627258777618} +{"time": "07:36:59", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5729404091835022, "noul_T": 0.9982354923377604, "noul_auc": 0.7482236586724276, "noul_acc": 0.7, "noul_f1": 0.7648902821316614, "noul_nll": 0.5743029379993677, "noul_ece": 0.03883609342575074, "noul_brier": 0.19543467934533204, "noul_sharp": 0.20428746938705444, "entropy": 0.573294460773468} +{"time": "07:40:32", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 844.8141875272004, "loss": 0.5964076686650515, "entropy": 0.5592931620776653, "sharp": 0.2207049708813429} +{"time": "07:43:56", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 846.0634813782683, "loss": 0.6241985633969307, "entropy": 0.5427652597427368, "sharp": 0.22682858258485794} +{"time": "07:47:18", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 847.2604479487751, "loss": 0.6075662449002266, "entropy": 0.5822915621101856, "sharp": 0.20644912961870432} +{"time": "07:50:44", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 848.2563489492785, "loss": 0.5968886781483889, "entropy": 0.5982044860720634, "sharp": 0.18787799216806889} +{"time": "07:54:26", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 849.1618582314821, "loss": 0.6160915717482567, "entropy": 0.638366349041462, "sharp": 0.1358610400930047} +{"time": "07:57:41", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 850.0349507502527, "loss": 0.5845605134963989, "entropy": 0.5795282498002052, "sharp": 0.19503102451562881} +{"time": "08:01:09", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 850.9235247627942, "loss": 0.6784631283953786, "entropy": 0.6191965397447348, "sharp": 0.15472826454788446} +{"time": "08:04:35", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 851.8718052081871, "loss": 0.641514134593308, "entropy": 0.6116195227950811, "sharp": 0.1695285625755787} +{"time": "08:08:07", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 852.8490080192149, "loss": 0.6352004371583462, "entropy": 0.5985357239842415, "sharp": 0.17505636252462864} +{"time": "08:11:37", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 853.8086374491934, "loss": 0.5996381919831038, "entropy": 0.5990317296236753, "sharp": 0.16090855933725834} +{"time": "08:13:01", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5792344212532043, "noul_T": 0.8294639878292215, "noul_auc": 0.7526286476177617, "noul_acc": 0.694, "noul_f1": 0.7520259319286872, "noul_nll": 0.5773319405019284, "noul_ece": 0.05647077721357346, "noul_brier": 0.19667747145139533, "noul_sharp": 0.1701449304819107, "entropy": 0.605682909488678} +{"time": "08:16:24", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 843.6170693590602, "loss": 0.5788559205830097, "entropy": 0.5862964540719986, "sharp": 0.1881586816161871} +{"time": "08:19:41", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 844.3131225216799, "loss": 0.6364476401358843, "entropy": 0.6155027598142624, "sharp": 0.16685261577367783} +{"time": "08:23:08", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 845.4478815719372, "loss": 0.5839954316616058, "entropy": 0.5760990530252457, "sharp": 0.19938300922513008} +{"time": "08:26:36", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 846.2431592117372, "loss": 0.6327791456133127, "entropy": 0.596554946154356, "sharp": 0.18539904616773129} +{"time": "08:29:56", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 846.9473084159663, "loss": 0.612697921693325, "entropy": 0.6151166297495365, "sharp": 0.15622410364449024} +{"time": "08:33:22", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 847.7998403097366, "loss": 0.6509111076593399, "entropy": 0.5906893685460091, "sharp": 0.19280972331762314} +{"time": "08:36:46", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 848.7406950331157, "loss": 0.6207835357636213, "entropy": 0.5798268960788846, "sharp": 0.19936123490333557} +{"time": "08:40:21", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 849.3848933393982, "loss": 0.6433960124850273, "entropy": 0.6129121631383896, "sharp": 0.16489386931061745} +{"time": "08:43:56", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 850.3392574781294, "loss": 0.6328387930989265, "entropy": 0.6129023134708405, "sharp": 0.15844745468348265} +{"time": "08:47:17", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 850.7003358151389, "loss": 0.606182437390089, "entropy": 0.5900630839169025, "sharp": 0.17659891489893198} +{"time": "08:48:41", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5694525837898254, "noul_T": 0.9609454538096077, "noul_auc": 0.7533712511181246, "noul_acc": 0.708, "noul_f1": 0.7760736196319018, "noul_nll": 0.5684477911368012, "noul_ece": 0.026868412375450134, "noul_brier": 0.19337661670677833, "noul_sharp": 0.1959773600101471, "entropy": 0.5763464570045471} +{"time": "08:52:06", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 843.1700941437473, "loss": 0.6088475957512856, "entropy": 0.6129441857337952, "sharp": 0.16532769612967968} +{"time": "08:55:23", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 843.7752599933468, "loss": 0.6615106035023928, "entropy": 0.6103065647184849, "sharp": 0.16539798490703106} +{"time": "08:58:55", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 844.5001691273674, "loss": 0.6171700805425644, "entropy": 0.6049262378364801, "sharp": 0.17070145532488823} +{"time": "09:02:17", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 845.2060976545603, "loss": 0.617370743304491, "entropy": 0.6071440801024437, "sharp": 0.16286395210772753} +{"time": "09:05:54", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 845.741471494207, "loss": 0.6169765461236238, "entropy": 0.6018830817192793, "sharp": 0.17968859616667032} +{"time": "09:09:13", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 846.2539482028287, "loss": 0.6045034937560558, "entropy": 0.6021101884543896, "sharp": 0.18125119619071484} +{"time": "09:12:33", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 846.8126621385611, "loss": 0.6471898816525936, "entropy": 0.6070581898093224, "sharp": 0.17797877453267574} +{"time": "09:16:09", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 847.4535006554576, "loss": 0.6092151775956154, "entropy": 0.5899638906121254, "sharp": 0.19304905831813812} +{"time": "09:19:42", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 848.0413964337264, "loss": 0.5859963558614254, "entropy": 0.5923516303300858, "sharp": 0.19085353799164295} +{"time": "09:23:17", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 848.6744031655705, "loss": 0.5692550614476204, "entropy": 0.5581959709525108, "sharp": 0.22975198924541473} +{"time": "09:24:41", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5732789635658264, "noul_T": 0.8397547860913183, "noul_auc": 0.7540800999139254, "noul_acc": 0.704, "noul_f1": 0.7658227848101266, "noul_nll": 0.5716236111819744, "noul_ece": 0.04152087658643723, "noul_brier": 0.1944882483364401, "noul_sharp": 0.1755858212709427, "entropy": 0.599066972732544} +{"time": "09:27:42", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9113372, "tok_per_s": 828.5527392816117, "rows_per_s": 1.5312098779377614, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5641881227493286, "noul_T": 0.8311502624434424, "choice_ce": 0.9396607279777527, "choice_T": 0.9575453686506029, "score_ce": 1.273746371269226, "score_T": 0.9968860395022641, "noul_auc": 0.7455445333333334, "noul_acc": 0.698, "noul_f1": 0.770516717325228, "noul_nll": 0.5714521871805192, "noul_ece": 0.03739325839281083, "noul_brier": 0.19463277056479883, "noul_sharp": 0.17946934700012207, "entropy": 0.5962491631507874}, "dropped_long": 0, "peak_gib": 18.237416744232178} diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/DONE b/e41/S1-E41-FRESH-A100-ts0-syn/DONE new file mode 100644 index 0000000000000000000000000000000000000000..f6be05944334a283fb68056ae602913cfe843540 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/DONE @@ -0,0 +1 @@ +2026-10-30 15:56:39 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/README.md b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/README.md new file mode 100644 index 0000000000000000000000000000000000000000..57167079ef3f181aacdbd678a733d71697d18f63 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-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 + + + + + +## Model Details + +### Model Description + + + + + +- **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] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +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 + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +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] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[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 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/adapter_config.json b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..9490d6212a4ddbabf91a6654b8efff0b990a7b8b --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-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": "0.21.2", + "qalora_group_size": 16, + "r": 16, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "k_proj", + "gate_proj", + "v_proj", + "up_proj", + "q_proj", + "out_proj", + "in_proj_z", + "in_proj_qkv", + "down_proj", + "o_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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9419652fa7f33e3973642d4b25f2c08d677fcafa --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38dce770a7485d58461f00e8b435981b8a59fa7b4d85698a39adcec19c72bb41 +size 160486456 diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/adapter_config.json b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..fadf1add9c5135762b687d4f5260deb938741ff5 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-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": "0.21.2", + "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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7124552aa7612074e965b394c47e0f7a11004468 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06c3779c253e4392f2eff35db457059eac54eb12d6a0489ebbec0a02334e76ca +size 176641832 diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json new file mode 100644 index 0000000000000000000000000000000000000000..fcc65e2631f2f7fcde05d1d2a339c357467e65b1 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json @@ -0,0 +1,99 @@ +{ + "bias": [ + -7.100164566509193e-06, + 7.09015967004234e-06, + -0.0009966813959181309, + -0.0005742351058870554, + 5.1083068683510646e-05, + 0.0013195601059123874, + 0.000959161901846528, + -0.0011933707864955068, + -0.000500809692312032, + -0.00012455931573640555, + -0.00045661081094294786, + -0.0003613382577896118, + -0.0011202807072550058, + -0.0005799551145173609, + -0.002972770482301712, + -0.0028353428933769464, + -0.0004124351544305682, + -0.0027276587206870317, + -0.002657216740772128, + -0.002570390235632658, + -0.002096695126965642, + -0.002580174943432212, + -0.0025975394528359175, + -0.0025782291777431965 + ], + "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" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/calibration.json b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/calibration.json new file mode 100644 index 0000000000000000000000000000000000000000..522524a51d2fbef3c70b3aabeebfc644db1c0a3a --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/calibration.json @@ -0,0 +1,29 @@ +{ + "version": 1, + "per_kind": { + "noul": 0.8549924184222403, + "choice": 0.9153525484025449, + "score": 1.0077701222049835 + }, + "fit": { + "noul": { + "n": 1087, + "T": 0.8549924184222403, + "ce_before": 0.5604386329650879, + "ce_after": 0.558404803276062 + }, + "choice": { + "n": 60, + "T": 0.9153525484025449, + "ce_before": 0.9069938659667969, + "ce_after": 0.904958963394165 + }, + "score": { + "n": 53, + "T": 1.0077701222049835, + "ce_before": 1.3230611085891724, + "ce_after": 1.323042869567871 + } + }, + "source": "models/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/head.safetensors b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/head.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2e6913e098378ba111ef33b29e922efb1523f0dd --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/head.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d80e0043a72b61a5e722ea6ab97f64f80771601b624b4734dac4efff21429771 +size 393464 diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/judge_config.json b/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000/judge_config.json new file mode 100644 index 0000000000000000000000000000000000000000..cc9d420148e800c4389f93589cd1af94622b848b --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-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/e41/S1-E41-FRESH-A100-ts0-syn/ckpt-16000", + "softcap": null, + "e16": { + "data": "data/e18/G-d512-n16k-a100", + "data_manifest": { + "arm": "G-d512-n16k-a100", + "name": null, + "train_sha256": "e68e935b88943ba54ed4296ecedd5f5b36a7b9f114cd4d203c4f67b6ff70b912", + "n_train": 16000 + }, + "init": "fresh", + "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", + "event": "ckpt", + "tag": "ckpt-16000", + "rows_seen": 16000, + "replay_seen": 842, + "step": 500, + "tokens": 9113372, + "tok_per_s": 828.8332566607069, + "rows_per_s": 1.531728289158018, + "val": { + "n": 1200, + "dropped_long": 0, + "noul_ce": 0.5604386329650879, + "noul_T": 0.8549924184222403, + "choice_ce": 0.9069938659667969, + "choice_T": 0.9153525484025449, + "score_ce": 1.3230611085891724, + "score_T": 1.0077701222049835, + "noul_auc": 0.7487274666666667, + "noul_acc": 0.704, + "noul_f1": 0.7764350453172205, + "noul_nll": 0.5678000621125102, + "noul_ece": 0.03950928713381291, + "noul_brier": 0.19362498969802922, + "noul_sharp": 0.18290674686431885, + "entropy": 0.5869004130363464 + }, + "dropped_long": 0, + "peak_gib": 18.237416744232178 + } +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts0-syn/train_log.jsonl b/e41/S1-E41-FRESH-A100-ts0-syn/train_log.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..d48af1a1b5cb12c34d46c008c068ad34464043ca --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts0-syn/train_log.jsonl @@ -0,0 +1,57 @@ +{"time": "12:53:22", "event": "start", "data": "data/e18/G-d512-n16k-a100", "data_manifest": {"arm": "G-d512-n16k-a100", "name": null, "train_sha256": "e68e935b88943ba54ed4296ecedd5f5b36a7b9f114cd4d203c4f67b6ff70b912", "n_train": 16000}, "init": "fresh", "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", "device": "cuda", "load_s": 68.57022666931152} +{"time": "12:56:51", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 863.6928242030459, "loss": 0.6713092941790819, "entropy": 0.4376097768545151, "sharp": 0.32057018019258976} +{"time": "13:00:20", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 866.472534477102, "loss": 0.6896154601126909, "entropy": 0.5052010118961334, "sharp": 0.2743905186653137} +{"time": "13:03:51", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 868.2846354715971, "loss": 0.6701861973851919, "entropy": 0.5078653134405613, "sharp": 0.26094710640609264} +{"time": "13:07:15", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 869.943572640944, "loss": 0.6415329053997993, "entropy": 0.654895655810833, "sharp": 0.11787931341677904} +{"time": "13:10:47", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 871.9250785426018, "loss": 0.5770723577588797, "entropy": 0.5072818398475647, "sharp": 0.25778891518712044} +{"time": "13:14:11", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 873.7121517002777, "loss": 0.6314391680061817, "entropy": 0.6818214803934097, "sharp": 0.059891607612371445} +{"time": "13:17:41", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 873.7017609180654, "loss": 0.6747958213090897, "entropy": 0.5531339049339294, "sharp": 0.22739075124263763} +{"time": "13:21:08", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 873.7113455549708, "loss": 0.8201661836355925, "entropy": 0.4519538311287761, "sharp": 0.31547821685671806} +{"time": "13:24:41", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 875.049233610559, "loss": 0.8304273784160614, "entropy": 0.5279915202409029, "sharp": 0.23154185991734266} +{"time": "13:28:11", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 874.7296253546453, "loss": 0.5787552297115326, "entropy": 0.6240838803350925, "sharp": 0.16088588815182447} +{"time": "13:29:35", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.5814749598503113, "noul_T": 0.9398173403689165, "noul_auc": 0.7366457950076792, "noul_acc": 0.702, "noul_f1": 0.783744557329463, "noul_nll": 0.5884761177748442, "noul_ece": 0.055264677405357364, "noul_brier": 0.2012616657323906, "noul_sharp": 0.18428945541381836, "entropy": 0.5915597081184387} +{"time": "13:33:00", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 844.8298714127784, "loss": 0.6438105292618275, "entropy": 0.5425056107342243, "sharp": 0.23382707498967648} +{"time": "13:36:38", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 848.2698361673039, "loss": 0.5873049162328243, "entropy": 0.589454784989357, "sharp": 0.19424969889223576} +{"time": "13:40:15", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 851.2799959666567, "loss": 0.5765495598316193, "entropy": 0.5597486309707165, "sharp": 0.22212236188352108} +{"time": "13:43:45", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 852.7341035603478, "loss": 0.6318581011146307, "entropy": 0.6558144465088844, "sharp": 0.11861468106508255} +{"time": "13:47:17", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 854.1923533776428, "loss": 0.6952907405793667, "entropy": 0.6589962467551231, "sharp": 0.09438893385231495} +{"time": "13:50:40", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 855.5337491099222, "loss": 0.6478397734463215, "entropy": 0.530195988714695, "sharp": 0.23498248867690563} +{"time": "13:54:11", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 856.8933844276953, "loss": 0.6134394071996212, "entropy": 0.5899704098701477, "sharp": 0.18354002572596073} +{"time": "13:57:39", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 858.1675045325949, "loss": 0.6200570613145828, "entropy": 0.5777353346347809, "sharp": 0.19606578722596169} +{"time": "14:01:08", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 858.8913339256663, "loss": 0.6576648037880659, "entropy": 0.5878043659031391, "sharp": 0.18177318386733532} +{"time": "14:04:37", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 859.8468043200365, "loss": 0.5590051729232073, "entropy": 0.5832923911511898, "sharp": 0.19537851959466934} +{"time": "14:06:00", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.574819803237915, "noul_T": 0.7946833053031614, "noul_auc": 0.7443924997046464, "noul_acc": 0.694, "noul_f1": 0.7642526964560863, "noul_nll": 0.577784072369337, "noul_ece": 0.04639053785800935, "noul_brier": 0.1974852471230014, "noul_sharp": 0.16745233535766602, "entropy": 0.6068589091300964} +{"time": "14:09:33", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 845.1291223590493, "loss": 0.5922418087720871, "entropy": 0.577846659347415, "sharp": 0.20129293017089367} +{"time": "14:12:58", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 846.3115382633157, "loss": 0.6056756228208542, "entropy": 0.5993921421468258, "sharp": 0.17416005581617355} +{"time": "14:16:20", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 847.4882848003472, "loss": 0.5529930107295513, "entropy": 0.5594154875725508, "sharp": 0.23192285746335983} +{"time": "14:19:46", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 848.5019235441844, "loss": 0.5733155775815248, "entropy": 0.5937597304582596, "sharp": 0.18758900836110115} +{"time": "14:23:28", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 849.3508985003341, "loss": 0.597817275673151, "entropy": 0.5590011738240719, "sharp": 0.20554181560873985} +{"time": "14:26:44", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 850.1948725386904, "loss": 0.5701239965856075, "entropy": 0.5964581519365311, "sharp": 0.17684556543827057} +{"time": "14:30:12", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 851.0679633021015, "loss": 0.7166153998114169, "entropy": 0.5814250502735376, "sharp": 0.1929884199053049} +{"time": "14:33:38", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 851.9920220641031, "loss": 0.6135260448791087, "entropy": 0.6097780596464872, "sharp": 0.16742566041648388} +{"time": "14:37:09", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 852.9757907713945, "loss": 0.6044531650841236, "entropy": 0.5505186505615711, "sharp": 0.2132402751594782} +{"time": "14:40:40", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 853.9143988486036, "loss": 0.5293492507189512, "entropy": 0.5522036626935005, "sharp": 0.19933302700519562} +{"time": "14:42:03", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5709512829780579, "noul_T": 1.0703478421113128, "noul_auc": 0.7472278948878499, "noul_acc": 0.692, "noul_f1": 0.7793696275071633, "noul_nll": 0.5785570668205618, "noul_ece": 0.06075401973724365, "noul_brier": 0.19784424995954902, "noul_sharp": 0.20544110238552094, "entropy": 0.558855414390564} +{"time": "14:45:26", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 843.8198811761192, "loss": 0.551854869350791, "entropy": 0.6008791755884886, "sharp": 0.1678249016404152} +{"time": "14:48:43", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 844.5285486457444, "loss": 0.645989534445107, "entropy": 0.5927655454725027, "sharp": 0.18859390541911125} +{"time": "14:52:10", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 845.6713923972628, "loss": 0.5708102434873581, "entropy": 0.5793188624083996, "sharp": 0.19270619936287403} +{"time": "14:55:37", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 846.5071688146895, "loss": 0.6277789641171694, "entropy": 0.5686078928411007, "sharp": 0.1970699280500412} +{"time": "14:58:57", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 847.2280999946047, "loss": 0.6118146330118179, "entropy": 0.6094939559698105, "sharp": 0.1570132439956069} +{"time": "15:02:22", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 848.1498778676824, "loss": 0.6264545992016792, "entropy": 0.5884319245815277, "sharp": 0.1982570542022586} +{"time": "15:05:46", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 849.1176968177975, "loss": 0.6122457422316074, "entropy": 0.5481799161061645, "sharp": 0.22755835391581059} +{"time": "15:09:20", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 849.8169978017618, "loss": 0.6056774444878101, "entropy": 0.5900398269295692, "sharp": 0.17994816228747368} +{"time": "15:12:55", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 850.7653178150629, "loss": 0.6357017587870359, "entropy": 0.5988152716308832, "sharp": 0.17185774724930525} +{"time": "15:16:16", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 851.172555224477, "loss": 0.5768326930701733, "entropy": 0.5895808264613152, "sharp": 0.17611087765544653} +{"time": "15:17:39", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5684843063354492, "noul_T": 0.9809907373033032, "noul_auc": 0.7492869318661288, "noul_acc": 0.698, "noul_f1": 0.7715582450832073, "noul_nll": 0.5735127909034491, "noul_ece": 0.05580830442905425, "noul_brier": 0.19552982967961646, "noul_sharp": 0.19600741565227509, "entropy": 0.5718139410018921} +{"time": "15:21:04", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 843.6534374309034, "loss": 0.59553462266922, "entropy": 0.5901413150131702, "sharp": 0.18474289868026972} +{"time": "15:24:22", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 844.2238761037845, "loss": 0.6868984177708626, "entropy": 0.601925287861377, "sharp": 0.16990244947373867} +{"time": "15:27:54", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 844.9026380223024, "loss": 0.5969959134235978, "entropy": 0.5690119271166623, "sharp": 0.1926587037742138} +{"time": "15:31:16", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 845.5582931844316, "loss": 0.6044984497129917, "entropy": 0.5929219871759415, "sharp": 0.17549285385757685} +{"time": "15:34:53", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 846.1139715472959, "loss": 0.5706863030791283, "entropy": 0.5779353454709053, "sharp": 0.20259064994752407} +{"time": "15:38:12", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 846.5962464791137, "loss": 0.584463506937027, "entropy": 0.58063486777246, "sharp": 0.1995541714131832} +{"time": "15:41:32", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 847.1615541904675, "loss": 0.5994970425963402, "entropy": 0.5794787369668484, "sharp": 0.20514125563204288} +{"time": "15:45:08", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 847.8001818506689, "loss": 0.5906099919229746, "entropy": 0.5592567771673203, "sharp": 0.22199173085391521} +{"time": "15:48:42", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 848.389760081328, "loss": 0.6159779727458954, "entropy": 0.5766282454133034, "sharp": 0.19679305516183376} +{"time": "15:52:16", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 849.0034778448423, "loss": 0.55549580976367, "entropy": 0.53367018699646, "sharp": 0.2412318717688322} +{"time": "15:53:40", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5694769024848938, "noul_T": 0.8777261662088477, "noul_auc": 0.7504514691735161, "noul_acc": 0.704, "noul_f1": 0.768025078369906, "noul_nll": 0.5725839892104269, "noul_ece": 0.04973726803064346, "noul_brier": 0.19539573194139032, "noul_sharp": 0.1799151748418808, "entropy": 0.5889390110969543} +{"time": "15:56:39", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9113372, "tok_per_s": 828.8332566607069, "rows_per_s": 1.531728289158018, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5604386329650879, "noul_T": 0.8549924184222403, "choice_ce": 0.9069938659667969, "choice_T": 0.9153525484025449, "score_ce": 1.3230611085891724, "score_T": 1.0077701222049835, "noul_auc": 0.7487274666666667, "noul_acc": 0.704, "noul_f1": 0.7764350453172205, "noul_nll": 0.5678000621125102, "noul_ece": 0.03950928713381291, "noul_brier": 0.19362498969802922, "noul_sharp": 0.18290674686431885, "entropy": 0.5869004130363464}, "dropped_long": 0, "peak_gib": 18.237416744232178} diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/DONE b/e41/S1-E41-FRESH-A100-ts1-syn/DONE new file mode 100644 index 0000000000000000000000000000000000000000..cd2d2ddc7f4c23e701ec638422276c12b301cd9d --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/DONE @@ -0,0 +1 @@ +2026-10-31 03:49:01 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/README.md b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/README.md new file mode 100644 index 0000000000000000000000000000000000000000..57167079ef3f181aacdbd678a733d71697d18f63 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-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 + + + + + +## Model Details + +### Model Description + + + + + +- **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] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +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 + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +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] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[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 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/adapter_config.json b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..fc6313e301f36c6e4fbac57d12c2a0967bfefbc2 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-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": "0.21.2", + "qalora_group_size": 16, + "r": 16, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "o_proj", + "q_proj", + "v_proj", + "out_proj", + "in_proj_qkv", + "in_proj_z", + "gate_proj", + "down_proj", + "k_proj", + "up_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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/adapter_model.safetensors b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..44c957a6c1754cc815fc564ffe3a4846a9e654be --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6143dbf17f7a54e4d7385dfa2d1cf244a88d8af98a85d45f81579b321f7ebc06 +size 160486456 diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/adapter_config.json b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..fadf1add9c5135762b687d4f5260deb938741ff5 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-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": "0.21.2", + "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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7476d99281a4a5e46219c1abd464bc001a82c080 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2d36335bc666c336f5f14eb56b5f153a1e9d25c33f1bb3a67f9e74bec7b2ae1 +size 176641832 diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/decision_head.json b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/decision_head.json new file mode 100644 index 0000000000000000000000000000000000000000..2e53997867bbdf8051c32686586512512a9058af --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/adapter_vllm/decision_head.json @@ -0,0 +1,99 @@ +{ + "bias": [ + 3.3104358863056405e-06, + -3.319411916891113e-06, + -0.000487992336275056, + -0.0007003077771514654, + -0.00017936660151463002, + 0.0009540123865008354, + 0.0007385392091237009, + -0.0010788452345877886, + -0.00048183323815464973, + -0.00027273784507997334, + -0.000723338860552758, + -0.00020596178364939988, + -0.00025717803509905934, + -0.0005367447156459093, + -0.003449787152931094, + -0.0033858707174658775, + -0.0007001151680015028, + -0.0029139157850295305, + -0.002948142820969224, + -0.002528570592403412, + -0.0022543014492839575, + -0.002792063169181347, + -0.0027894279919564724, + -0.002860283013433218 + ], + "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" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/calibration.json b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/calibration.json new file mode 100644 index 0000000000000000000000000000000000000000..8e40ef6e6601dfd5318e9fc4a07d685e2475b1b2 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/calibration.json @@ -0,0 +1,29 @@ +{ + "version": 1, + "per_kind": { + "noul": 0.955422489539055, + "choice": 0.8909501705556426, + "score": 1.0035458198119358 + }, + "fit": { + "noul": { + "n": 1087, + "T": 0.955422489539055, + "ce_before": 0.5597464442253113, + "ce_after": 0.5595545768737793 + }, + "choice": { + "n": 60, + "T": 0.8909501705556426, + "ce_before": 0.9309567809104919, + "ce_after": 0.9275316596031189 + }, + "score": { + "n": 53, + "T": 1.0035458198119358, + "ce_before": 1.2876887321472168, + "ce_after": 1.287685513496399 + } + }, + "source": "models/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/head.safetensors b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/head.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dbb20afa337b0c1684aa1c624e780a02965afb49 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/head.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4babd4f7efef33e1b46f72c8b42df92abf52c7bed9b6aadcce0a3f97239a2f45 +size 393464 diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/judge_config.json b/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000/judge_config.json new file mode 100644 index 0000000000000000000000000000000000000000..bed9ef32c90dbed28656066e1a2fde159b546f94 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-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/e41/S1-E41-FRESH-A100-ts1-syn/ckpt-16000", + "softcap": null, + "e16": { + "data": "data/e18/G-d512-n16k-a100", + "data_manifest": { + "arm": "G-d512-n16k-a100", + "name": null, + "train_sha256": "e68e935b88943ba54ed4296ecedd5f5b36a7b9f114cd4d203c4f67b6ff70b912", + "n_train": 16000 + }, + "init": "fresh", + "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": 1, + "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", + "event": "ckpt", + "tag": "ckpt-16000", + "rows_seen": 16000, + "replay_seen": 842, + "step": 500, + "tokens": 9113372, + "tok_per_s": 777.7728396930712, + "rows_per_s": 1.437365901503131, + "val": { + "n": 1200, + "dropped_long": 0, + "noul_ce": 0.5597464442253113, + "noul_T": 0.955422489539055, + "choice_ce": 0.9309567809104919, + "choice_T": 0.8909501705556426, + "score_ce": 1.2876887321472168, + "score_T": 1.0035458198119358, + "noul_auc": 0.7458645333333334, + "noul_acc": 0.703, + "noul_f1": 0.7741444866920152, + "noul_nll": 0.5702573186308145, + "noul_ece": 0.027028134137392044, + "noul_brier": 0.1945888024108201, + "noul_sharp": 0.20050957798957825, + "entropy": 0.5704321265220642 + }, + "dropped_long": 0, + "peak_gib": 18.23796796798706 + } +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100-ts1-syn/train_log.jsonl b/e41/S1-E41-FRESH-A100-ts1-syn/train_log.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..b0496f1e6edfc28d3a12d2dca2b130fe09647290 --- /dev/null +++ b/e41/S1-E41-FRESH-A100-ts1-syn/train_log.jsonl @@ -0,0 +1,57 @@ +{"time": "00:33:42", "event": "start", "data": "data/e18/G-d512-n16k-a100", "data_manifest": {"arm": "G-d512-n16k-a100", "name": null, "train_sha256": "e68e935b88943ba54ed4296ecedd5f5b36a7b9f114cd4d203c4f67b6ff70b912", "n_train": 16000}, "init": "fresh", "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": 1, "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", "device": "cuda", "load_s": 73.01242423057556} +{"time": "00:37:17", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 801.7353781082578, "loss": 0.625093150883913, "entropy": 0.5605938360095024, "sharp": 0.20331614464521408} +{"time": "00:41:04", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 811.8554971777578, "loss": 0.6884083030745387, "entropy": 0.6013521924614906, "sharp": 0.17694772314280272} +{"time": "00:44:47", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 816.0051883601197, "loss": 0.8310562521219254, "entropy": 0.6221914812922478, "sharp": 0.15975451935082674} +{"time": "00:48:38", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 817.7647111395037, "loss": 0.6669170372188091, "entropy": 0.6176962852478027, "sharp": 0.15694124717265368} +{"time": "00:52:22", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 820.1057563521265, "loss": 0.7178054396063089, "entropy": 0.6445112843066454, "sharp": 0.1272907890379429} +{"time": "00:56:08", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 821.4990930038683, "loss": 0.7376941964030266, "entropy": 0.4915303271263838, "sharp": 0.25296232663095} +{"time": "00:59:37", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 822.4873250820289, "loss": 0.6361687332391739, "entropy": 0.5918503850698471, "sharp": 0.1824391083791852} +{"time": "01:03:08", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 821.3465662481968, "loss": 0.6817021518945694, "entropy": 0.6088520549237728, "sharp": 0.17248315922915936} +{"time": "01:06:55", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 822.22463769497, "loss": 0.663878446444869, "entropy": 0.542602926492691, "sharp": 0.22113246656954288} +{"time": "01:10:37", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 822.4432603599707, "loss": 0.6110749542713165, "entropy": 0.6543462052941322, "sharp": 0.10466349124908447} +{"time": "01:12:06", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.594371497631073, "noul_T": 0.7058961675100118, "noul_auc": 0.7087644090395099, "noul_acc": 0.662, "noul_f1": 0.7363494539781591, "noul_nll": 0.6041911796182394, "noul_ece": 0.04898825371265412, "noul_brier": 0.2086251450180036, "noul_sharp": 0.1410246193408966, "entropy": 0.6325206160545349} +{"time": "01:15:48", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 793.7972428929481, "loss": 0.6050500255078077, "entropy": 0.6188328266143799, "sharp": 0.1569659160450101} +{"time": "01:19:29", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 795.876896358572, "loss": 0.6076511926949024, "entropy": 0.5218279547989368, "sharp": 0.2424597144126892} +{"time": "01:23:13", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 798.3937947228087, "loss": 0.5421918649226427, "entropy": 0.560256464406848, "sharp": 0.21816709451377392} +{"time": "01:26:38", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 799.0711963504448, "loss": 0.6151688061654568, "entropy": 0.5798659473657608, "sharp": 0.20298618637025356} +{"time": "01:30:21", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 800.4613219520712, "loss": 0.6030701696872711, "entropy": 0.570500323548913, "sharp": 0.2114113848656416} +{"time": "01:34:07", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 801.3304622308992, "loss": 0.7395098991692066, "entropy": 0.5972349531948566, "sharp": 0.18211051728576422} +{"time": "01:37:57", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 802.853889232634, "loss": 0.5469484739005566, "entropy": 0.5435362607240677, "sharp": 0.2340071927756071} +{"time": "01:41:35", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 804.0239722681453, "loss": 0.6287239696830511, "entropy": 0.5381829086691141, "sharp": 0.22116632014513016} +{"time": "01:45:20", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 804.4293969218749, "loss": 0.6560296714305878, "entropy": 0.5701234303414822, "sharp": 0.20820186380296946} +{"time": "01:49:06", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 805.3622559463774, "loss": 0.6167759075760841, "entropy": 0.5404207184910774, "sharp": 0.2274984922260046} +{"time": "01:50:36", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5811861753463745, "noul_T": 1.3247820690239431, "noul_auc": 0.7425359909537391, "noul_acc": 0.692, "noul_f1": 0.7708333333333334, "noul_nll": 0.5916286239847541, "noul_ece": 0.07714591783285141, "noul_brier": 0.201104718821847, "noul_sharp": 0.2425694316625595, "entropy": 0.515974223613739} +{"time": "01:54:15", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 790.9798751926415, "loss": 0.6386554390192032, "entropy": 0.6249810867011547, "sharp": 0.1498854784294963} +{"time": "01:57:54", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 791.8848516795925, "loss": 0.6112443543970585, "entropy": 0.6071700379252434, "sharp": 0.1820289958268404} +{"time": "02:01:35", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 793.2052645799918, "loss": 0.6147701535373926, "entropy": 0.5441010240465403, "sharp": 0.23971235286444426} +{"time": "02:05:15", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 794.1512574179088, "loss": 0.6314794346690178, "entropy": 0.5643994025886059, "sharp": 0.210531584918499} +{"time": "02:08:59", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 795.1211585410332, "loss": 0.5746912751346827, "entropy": 0.5303062852472067, "sharp": 0.24345698859542608} +{"time": "02:12:48", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 796.6057566756658, "loss": 0.5731548089534044, "entropy": 0.6093357168138027, "sharp": 0.15982440300285816} +{"time": "02:16:24", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 796.9783535881069, "loss": 0.6233068555593491, "entropy": 0.594236746430397, "sharp": 0.1817496307194233} +{"time": "02:20:09", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 797.6295773579889, "loss": 0.6718552969396114, "entropy": 0.5666984394192696, "sharp": 0.19890718813985586} +{"time": "02:23:47", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 798.4048070622127, "loss": 0.6132985129952431, "entropy": 0.6179444007575512, "sharp": 0.15793745405972004} +{"time": "02:27:22", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 799.2563910090082, "loss": 0.6461672186851501, "entropy": 0.5662364400923252, "sharp": 0.20489921979606152} +{"time": "02:28:52", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5671845078468323, "noul_T": 0.980271842707642, "noul_auc": 0.7565610706992286, "noul_acc": 0.7, "noul_f1": 0.7706422018348624, "noul_nll": 0.5662861945852637, "noul_ece": 0.033144451498985296, "noul_brier": 0.1924687937072352, "noul_sharp": 0.20048245787620544, "entropy": 0.5708160996437073} +{"time": "02:32:35", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 789.7942049438942, "loss": 0.6109202988445759, "entropy": 0.5839223265647888, "sharp": 0.19965385645627975} +{"time": "02:36:09", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 790.9111340690996, "loss": 0.5538351722061634, "entropy": 0.548102755099535, "sharp": 0.21408056002110243} +{"time": "02:39:49", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 792.0010401930376, "loss": 0.5358663443475962, "entropy": 0.5421231277287006, "sharp": 0.21509743481874466} +{"time": "02:43:39", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 793.0394142314675, "loss": 0.6147301644086838, "entropy": 0.5921670943498611, "sharp": 0.18138292618095875} +{"time": "02:47:19", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 793.9938331840793, "loss": 0.5278559774160385, "entropy": 0.5549597144126892, "sharp": 0.21428301464766264} +{"time": "02:50:59", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 794.961751697438, "loss": 0.6752193085849285, "entropy": 0.5887222215533257, "sharp": 0.1798875518143177} +{"time": "02:54:30", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 795.5895600552515, "loss": 0.563351908698678, "entropy": 0.57802489772439, "sharp": 0.20253428351134062} +{"time": "02:58:09", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 796.4353184723622, "loss": 0.6468511838465929, "entropy": 0.5910973027348518, "sharp": 0.18465449009090662} +{"time": "03:01:45", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 797.1222715104967, "loss": 0.674864262342453, "entropy": 0.6051372364163399, "sharp": 0.16976209077984095} +{"time": "03:05:28", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 797.5574325516012, "loss": 0.549790520220995, "entropy": 0.5744068995118141, "sharp": 0.20274406857788563} +{"time": "03:06:58", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.565473735332489, "noul_T": 0.9793171551577307, "noul_auc": 0.7529661946633812, "noul_acc": 0.704, "noul_f1": 0.7797619047619048, "noul_nll": 0.5694120173305273, "noul_ece": 0.03718629491329194, "noul_brier": 0.19405227831912145, "noul_sharp": 0.20141594111919403, "entropy": 0.5692461133003235} +{"time": "03:10:36", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 790.4422014678283, "loss": 0.5917508453130722, "entropy": 0.592524953186512, "sharp": 0.18488841876387596} +{"time": "03:14:27", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 791.1326645506989, "loss": 0.5730487797409296, "entropy": 0.5778018459677696, "sharp": 0.20740890130400658} +{"time": "03:18:10", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 791.9829030923853, "loss": 0.5902607813477516, "entropy": 0.5705467090010643, "sharp": 0.19985929317772388} +{"time": "03:22:01", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 792.9812182553476, "loss": 0.6773195117712021, "entropy": 0.5950019359588623, "sharp": 0.18867509346455336} +{"time": "03:25:47", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 793.6735820964622, "loss": 0.6417971774935722, "entropy": 0.6100424500182271, "sharp": 0.16594750992953777} +{"time": "03:29:31", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 794.1549652942956, "loss": 0.5897891446948051, "entropy": 0.5865767523646355, "sharp": 0.19172492064535618} +{"time": "03:33:16", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 794.7613327529679, "loss": 0.6191605981439352, "entropy": 0.5717110354453325, "sharp": 0.20864122174680233} +{"time": "03:36:55", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 795.44543641072, "loss": 0.6006900761276484, "entropy": 0.5867863334715366, "sharp": 0.1947379419580102} +{"time": "03:40:41", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 796.0700713389465, "loss": 0.576278068125248, "entropy": 0.5964784696698189, "sharp": 0.18708810210227966} +{"time": "03:44:20", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 796.7743833411222, "loss": 0.5446002520620823, "entropy": 0.5656477250158787, "sharp": 0.2000382924452424} +{"time": "03:45:49", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.568437933921814, "noul_T": 0.9912234223446877, "noul_auc": 0.7504008371166732, "noul_acc": 0.706, "noul_f1": 0.7692307692307693, "noul_nll": 0.572175752684474, "noul_ece": 0.052468656718730924, "noul_brier": 0.19493022387134773, "noul_sharp": 0.19993163645267487, "entropy": 0.570095956325531} +{"time": "03:49:01", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 842, "step": 500, "tokens": 9113372, "tok_per_s": 777.7728396930712, "rows_per_s": 1.437365901503131, "val": {"n": 1200, "dropped_long": 0, "noul_ce": 0.5597464442253113, "noul_T": 0.955422489539055, "choice_ce": 0.9309567809104919, "choice_T": 0.8909501705556426, "score_ce": 1.2876887321472168, "score_T": 1.0035458198119358, "noul_auc": 0.7458645333333334, "noul_acc": 0.703, "noul_f1": 0.7741444866920152, "noul_nll": 0.5702573186308145, "noul_ece": 0.027028134137392044, "noul_brier": 0.1945888024108201, "noul_sharp": 0.20050957798957825, "entropy": 0.5704321265220642}, "dropped_long": 0, "peak_gib": 18.23796796798706} diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/DONE b/e41/S1-E41-FRESH-A100NR-ts0-syn/DONE new file mode 100644 index 0000000000000000000000000000000000000000..5165e0f787645462dd6eb279c841ce963253589d --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/DONE @@ -0,0 +1 @@ +2026-10-31 12:30:27 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/README.md b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/README.md new file mode 100644 index 0000000000000000000000000000000000000000..57167079ef3f181aacdbd678a733d71697d18f63 --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-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 + + + + + +## Model Details + +### Model Description + + + + + +- **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] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +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 + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +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] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[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 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/adapter_config.json b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..f5285757fcb867d679bf306bc1cc2e853019def7 --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-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": "0.21.2", + "qalora_group_size": 16, + "r": 16, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "down_proj", + "q_proj", + "out_proj", + "o_proj", + "k_proj", + "gate_proj", + "up_proj", + "in_proj_qkv", + "in_proj_z", + "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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7f02049e72d4ec2f2bc2785e608095f82f17385e --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9aaa5b1631264c2f17e9ecf2e5a39a45d6dd5060f565ae63aec5098831b62f8e +size 160486456 diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/adapter_config.json b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..fadf1add9c5135762b687d4f5260deb938741ff5 --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-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": "0.21.2", + "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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2c91a1a85f394db45f757a1447f064b1a0777222 --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a619b093b3ea22851974950b370f73c486db66e3d0c176671f148db3788df47 +size 176641832 diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json new file mode 100644 index 0000000000000000000000000000000000000000..885d5ef18febd5eb92571437d2f3958defe56f5f --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/adapter_vllm/decision_head.json @@ -0,0 +1,99 @@ +{ + "bias": [ + -6.92921967129223e-05, + 6.927984213689342e-05, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "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" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/calibration.json b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/calibration.json new file mode 100644 index 0000000000000000000000000000000000000000..7bc269c5b97127d603261efca9971604655fc703 --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/calibration.json @@ -0,0 +1,17 @@ +{ + "version": 1, + "per_kind": { + "noul": 0.8675597586182227, + "choice": 1.0, + "score": 1.0 + }, + "fit": { + "noul": { + "n": 1000, + "T": 0.8675597586182227, + "ce_before": 0.5639619827270508, + "ce_after": 0.5623140931129456 + } + }, + "source": "models/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/head.safetensors b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/head.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9606f5eb55c193a2892b4f4ebb340f9aa419c58d --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/head.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20c20650bd4585e5dcc0965ddb395d60cb0421406a21e720ff84d094b8db10eb +size 393464 diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/judge_config.json b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/judge_config.json new file mode 100644 index 0000000000000000000000000000000000000000..490bbdb17c7be55c3123f39ea424cd9feb37ec7a --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000/judge_config.json @@ -0,0 +1,160 @@ +{ + "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/e41/S1-E41-FRESH-A100NR-ts0-syn/ckpt-16000", + "softcap": null, + "e16": { + "data": "data/e18/G-d512-n16k-a100", + "data_manifest": { + "arm": "G-d512-n16k-a100", + "name": null, + "train_sha256": "e68e935b88943ba54ed4296ecedd5f5b36a7b9f114cd4d203c4f67b6ff70b912", + "n_train": 16000 + }, + "init": "fresh", + "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": null, + "replay_frac": 0.0, + "replay_sha256": null, + "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", + "event": "ckpt", + "tag": "ckpt-16000", + "rows_seen": 16000, + "replay_seen": 0, + "step": 500, + "tokens": 9001452, + "tok_per_s": 835.8174193861242, + "rows_per_s": 1.4856579476245457, + "val": { + "n": 1000, + "dropped_long": 0, + "noul_ce": 0.5639619827270508, + "noul_T": 0.8675597586182227, + "noul_auc": 0.7489578666666666, + "noul_acc": 0.704, + "noul_f1": 0.7767722473604827, + "noul_nll": 0.5669560533519834, + "noul_ece": 0.03547174583375453, + "noul_brier": 0.1934193399194031, + "noul_sharp": 0.18415376543998718, + "entropy": 0.5859837532043457 + }, + "dropped_long": 0, + "peak_gib": 18.23741626739502 + } +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-A100NR-ts0-syn/train_log.jsonl b/e41/S1-E41-FRESH-A100NR-ts0-syn/train_log.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..966e6923414ba6791dbc1e0cbd2e2a348baf4ef0 --- /dev/null +++ b/e41/S1-E41-FRESH-A100NR-ts0-syn/train_log.jsonl @@ -0,0 +1,57 @@ +{"time": "09:30:56", "event": "start", "data": "data/e18/G-d512-n16k-a100", "data_manifest": {"arm": "G-d512-n16k-a100", "name": null, "train_sha256": "e68e935b88943ba54ed4296ecedd5f5b36a7b9f114cd4d203c4f67b6ff70b912", "n_train": 16000}, "init": "fresh", "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": null, "replay_frac": 0.0, "replay_sha256": null, "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": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3", "device": "cuda", "load_s": 75.86485147476196} +{"time": "09:34:22", "event": "step", "step": 10, "rows_seen": 320, "lr": 7.333333333333333e-05, "tok_per_s": 859.8716677573857, "loss": 0.6354091130197048, "entropy": 0.45976495556533337, "sharp": 0.3083764370530844} +{"time": "09:37:48", "event": "step", "step": 20, "rows_seen": 640, "lr": 9.997508952966197e-05, "tok_per_s": 865.6396562382371, "loss": 0.8660826086997986, "entropy": 0.3398577943444252, "sharp": 0.36882008612155914} +{"time": "09:41:16", "event": "step", "step": 30, "rows_seen": 960, "lr": 9.977596250541417e-05, "tok_per_s": 870.2762467725187, "loss": 0.6395208984613419, "entropy": 0.5155012998729944, "sharp": 0.23681509494781494} +{"time": "09:44:35", "event": "step", "step": 40, "rows_seen": 1280, "lr": 9.937854366485982e-05, "tok_per_s": 874.554940005879, "loss": 0.601106621325016, "entropy": 0.6441787555813789, "sharp": 0.11436268594115973} +{"time": "09:48:03", "event": "step", "step": 50, "rows_seen": 1600, "lr": 9.8784499920729e-05, "tok_per_s": 876.1809450093167, "loss": 0.5056453384459019, "entropy": 0.5827620886266232, "sharp": 0.20102746225893497} +{"time": "09:51:24", "event": "step", "step": 60, "rows_seen": 1920, "lr": 9.79963228989294e-05, "tok_per_s": 876.9949859821481, "loss": 0.6350937634706497, "entropy": 0.6432440206408501, "sharp": 0.1235008854418993} +{"time": "09:54:49", "event": "step", "step": 70, "rows_seen": 2240, "lr": 9.701731848780134e-05, "tok_per_s": 878.1442819368517, "loss": 0.5880141071975231, "entropy": 0.6145525649189949, "sharp": 0.16659976355731487} +{"time": "09:58:14", "event": "step", "step": 80, "rows_seen": 2560, "lr": 9.585159297207315e-05, "tok_per_s": 878.0893623273032, "loss": 0.6189858019351959, "entropy": 0.6363483965396881, "sharp": 0.14846031740307808} +{"time": "10:01:42", "event": "step", "step": 90, "rows_seen": 2880, "lr": 9.450403580967612e-05, "tok_per_s": 879.7256697902923, "loss": 0.6234800703823566, "entropy": 0.6108531206846237, "sharp": 0.16964469477534294} +{"time": "10:05:08", "event": "step", "step": 100, "rows_seen": 3200, "lr": 9.298029912365855e-05, "tok_per_s": 879.9150925382897, "loss": 0.6143936812877655, "entropy": 0.6511763967573643, "sharp": 0.11758880782872438} +{"time": "10:06:31", "event": "val", "step": 100, "rows_seen": 3200, "n": 500, "dropped_long": 0, "noul_ce": 0.6092418432235718, "noul_T": 0.971464837155799, "noul_auc": 0.7238358846264199, "noul_acc": 0.682, "noul_f1": 0.7234782608695652, "noul_nll": 0.6123800744116307, "noul_ece": 0.10942058801651, "noul_brier": 0.21209673789207895, "noul_sharp": 0.16415363550186157, "entropy": 0.6130161881446838} +{"time": "10:09:54", "event": "step", "step": 110, "rows_seen": 3520, "lr": 9.128677399521752e-05, "tok_per_s": 848.8292261137519, "loss": 0.615717776119709, "entropy": 0.6158006228506565, "sharp": 0.16641773469746113} +{"time": "10:13:28", "event": "step", "step": 120, "rows_seen": 3840, "lr": 8.943056365728335e-05, "tok_per_s": 852.641210046723, "loss": 0.5958198010921478, "entropy": 0.6018266975879669, "sharp": 0.18515735678374767} +{"time": "10:17:01", "event": "step", "step": 130, "rows_seen": 4160, "lr": 8.741945370109202e-05, "tok_per_s": 856.0238365395676, "loss": 0.6094316616654396, "entropy": 0.6555486395955086, "sharp": 0.11222152598202229} +{"time": "10:20:25", "event": "step", "step": 140, "rows_seen": 4480, "lr": 8.526187942070997e-05, "tok_per_s": 857.8628012983557, "loss": 0.8630004338920116, "entropy": 0.5439274106174707, "sharp": 0.2181365992873907} +{"time": "10:23:51", "event": "step", "step": 150, "rows_seen": 4800, "lr": 8.296689043247915e-05, "tok_per_s": 859.9757642359211, "loss": 0.7891700081527233, "entropy": 0.5123810656368732, "sharp": 0.26871342957019806} +{"time": "10:27:12", "event": "step", "step": 160, "rows_seen": 5120, "lr": 8.054411271778132e-05, "tok_per_s": 861.3555967860905, "loss": 0.596333883702755, "entropy": 0.573711734265089, "sharp": 0.18731662072241306} +{"time": "10:30:38", "event": "step", "step": 170, "rows_seen": 5440, "lr": 7.800370824832674e-05, "tok_per_s": 862.805250029445, "loss": 0.6432756278663874, "entropy": 0.6450665295124054, "sharp": 0.1262973789125681} +{"time": "10:34:02", "event": "step", "step": 180, "rows_seen": 5760, "lr": 7.535633236331324e-05, "tok_per_s": 864.0888208264142, "loss": 0.6388018876314163, "entropy": 0.5693687871098518, "sharp": 0.2116860169917345} +{"time": "10:37:26", "event": "step", "step": 190, "rows_seen": 6080, "lr": 7.26130890772308e-05, "tok_per_s": 865.119798501627, "loss": 0.6288625970482826, "entropy": 0.5864464342594147, "sharp": 0.18553094938397408} +{"time": "10:40:51", "event": "step", "step": 200, "rows_seen": 6400, "lr": 6.978548450576568e-05, "tok_per_s": 866.0606169634237, "loss": 0.5659723617136478, "entropy": 0.5954342093318701, "sharp": 0.1942771291360259} +{"time": "10:42:14", "event": "val", "step": 200, "rows_seen": 6400, "n": 500, "dropped_long": 0, "noul_ce": 0.5779189467430115, "noul_T": 0.8567456254386052, "noul_auc": 0.736713304416803, "noul_acc": 0.708, "noul_f1": 0.7652733118971061, "noul_nll": 0.5831517841815949, "noul_ece": 0.06773575913906098, "noul_brier": 0.1994085176293151, "noul_sharp": 0.17612969875335693, "entropy": 0.6009074449539185} +{"time": "10:45:42", "event": "step", "step": 210, "rows_seen": 6720, "lr": 6.688537860515266e-05, "tok_per_s": 851.0965685886879, "loss": 0.5136166512966156, "entropy": 0.5459015090018511, "sharp": 0.22268066927790642} +{"time": "10:49:04", "event": "step", "step": 220, "rows_seen": 7040, "lr": 6.392493542739706e-05, "tok_per_s": 852.1583800430171, "loss": 0.6232382170855999, "entropy": 0.5539784021675587, "sharp": 0.21410788968205452} +{"time": "10:52:21", "event": "step", "step": 230, "rows_seen": 7360, "lr": 6.091657210001309e-05, "tok_per_s": 853.4093556778491, "loss": 0.5561170168220997, "entropy": 0.5499005280435085, "sharp": 0.23852031491696835} +{"time": "10:55:41", "event": "step", "step": 240, "rows_seen": 7680, "lr": 5.7872906744276015e-05, "tok_per_s": 854.9092340611226, "loss": 0.5586049221456051, "entropy": 0.5969101004302502, "sharp": 0.18936886452138424} +{"time": "10:59:17", "event": "step", "step": 250, "rows_seen": 8000, "lr": 5.48067055504363e-05, "tok_per_s": 855.9939361965822, "loss": 0.6076696440577507, "entropy": 0.5663048662245274, "sharp": 0.2031424380838871} +{"time": "11:02:31", "event": "step", "step": 260, "rows_seen": 8320, "lr": 5.173082923188119e-05, "tok_per_s": 856.6028435228961, "loss": 0.5831205435097218, "entropy": 0.6234708465635777, "sharp": 0.14855318423360586} +{"time": "11:05:53", "event": "step", "step": 270, "rows_seen": 8640, "lr": 4.865817908283331e-05, "tok_per_s": 857.8279023608127, "loss": 0.667628463357687, "entropy": 0.5734901046380401, "sharp": 0.19734588637948036} +{"time": "11:09:14", "event": "step", "step": 280, "rows_seen": 8960, "lr": 4.560164286583922e-05, "tok_per_s": 858.8332721185225, "loss": 0.5809339471161366, "entropy": 0.574310046620667, "sharp": 0.19706436805427074} +{"time": "11:12:42", "event": "step", "step": 290, "rows_seen": 9280, "lr": 4.257404075601443e-05, "tok_per_s": 859.7688850898597, "loss": 0.5717178024351597, "entropy": 0.5891993679106236, "sharp": 0.18384019751101732} +{"time": "11:16:06", "event": "step", "step": 300, "rows_seen": 9600, "lr": 3.958807156877371e-05, "tok_per_s": 860.9299139531589, "loss": 0.49903257796540856, "entropy": 0.5674651837907732, "sharp": 0.1861200165003538} +{"time": "11:17:30", "event": "val", "step": 300, "rows_seen": 9600, "n": 500, "dropped_long": 0, "noul_ce": 0.5679372549057007, "noul_T": 0.922112435802677, "noul_auc": 0.7519535535265227, "noul_acc": 0.706, "noul_f1": 0.776255707762557, "noul_nll": 0.570399238370359, "noul_ece": 0.05298358249664307, "noul_brier": 0.19426175436281728, "noul_sharp": 0.1896880716085434, "entropy": 0.5810537338256836} +{"time": "11:20:47", "event": "step", "step": 310, "rows_seen": 9920, "lr": 3.665625949658603e-05, "tok_per_s": 850.7499046245407, "loss": 0.566287849098444, "entropy": 0.595051484182477, "sharp": 0.17188620753586292} +{"time": "11:24:00", "event": "step", "step": 320, "rows_seen": 10240, "lr": 3.379090157815887e-05, "tok_per_s": 851.4701403242618, "loss": 0.6487684152089059, "entropy": 0.5856913579627872, "sharp": 0.19476856477558613} +{"time": "11:27:22", "event": "step", "step": 330, "rows_seen": 10560, "lr": 3.10040161203844e-05, "tok_per_s": 852.6034505017728, "loss": 0.5673067681491375, "entropy": 0.5588213875889778, "sharp": 0.21688863076269627} +{"time": "11:30:46", "event": "step", "step": 340, "rows_seen": 10880, "lr": 2.8307292289383776e-05, "tok_per_s": 853.4795500606288, "loss": 0.6145714931190014, "entropy": 0.5744422115385532, "sharp": 0.1963830664753914} +{"time": "11:34:02", "event": "step", "step": 350, "rows_seen": 11200, "lr": 2.571204108208194e-05, "tok_per_s": 854.0732513100978, "loss": 0.6039040610194206, "entropy": 0.6178362816572189, "sharp": 0.14937147311866283} +{"time": "11:37:23", "event": "step", "step": 360, "rows_seen": 11520, "lr": 2.322914788395595e-05, "tok_per_s": 854.9197465472693, "loss": 0.5668587535619736, "entropy": 0.5684940963983536, "sharp": 0.2162558389827609} +{"time": "11:40:42", "event": "step", "step": 370, "rows_seen": 11840, "lr": 2.0869026811945157e-05, "tok_per_s": 855.7657557762418, "loss": 0.6101383306086063, "entropy": 0.5523163303732872, "sharp": 0.22743568941950798} +{"time": "11:44:13", "event": "step", "step": 380, "rows_seen": 12160, "lr": 1.8641577034025627e-05, "tok_per_s": 856.4950421450243, "loss": 0.5764838438481092, "entropy": 0.569079365581274, "sharp": 0.19873671233654022} +{"time": "11:47:44", "event": "step", "step": 390, "rows_seen": 12480, "lr": 1.6556141248659472e-05, "tok_per_s": 857.2730330949901, "loss": 0.6161915138363838, "entropy": 0.597283560782671, "sharp": 0.1642635576426983} +{"time": "11:51:02", "event": "step", "step": 400, "rows_seen": 12800, "lr": 1.4621466498270773e-05, "tok_per_s": 857.5557359399667, "loss": 0.5721729546785355, "entropy": 0.5930621586740017, "sharp": 0.17312102485448122} +{"time": "11:52:25", "event": "val", "step": 400, "rows_seen": 12800, "n": 500, "dropped_long": 0, "noul_ce": 0.5700386762619019, "noul_T": 0.9651625422749267, "noul_auc": 0.7522235911630184, "noul_acc": 0.692, "noul_f1": 0.7761627906976745, "noul_nll": 0.573307738892734, "noul_ece": 0.05151889932155608, "noul_brier": 0.19593086339102336, "noul_sharp": 0.19447061419487, "entropy": 0.5760045051574707} +{"time": "11:55:46", "event": "step", "step": 410, "rows_seen": 13120, "lr": 1.284566748110963e-05, "tok_per_s": 849.8084877970246, "loss": 0.5977747403085232, "entropy": 0.5986251309514046, "sharp": 0.17813397478312254} +{"time": "11:58:58", "event": "step", "step": 420, "rows_seen": 13440, "lr": 1.1236192515387821e-05, "tok_per_s": 850.5725996451521, "loss": 0.6522680856287479, "entropy": 0.5997651382349432, "sharp": 0.17166856862604618} +{"time": "12:02:26", "event": "step", "step": 430, "rows_seen": 13760, "lr": 9.799792298443728e-06, "tok_per_s": 851.2455578701472, "loss": 0.5759519934654236, "entropy": 0.5718252919614315, "sharp": 0.1896313987672329} +{"time": "12:05:43", "event": "step", "step": 440, "rows_seen": 14080, "lr": 8.542491591971554e-06, "tok_per_s": 851.9945473657621, "loss": 0.5819804780185223, "entropy": 0.6032476760447025, "sharp": 0.1678545894101262} +{"time": "12:09:16", "event": "step", "step": 450, "rows_seen": 14400, "lr": 7.469563952077155e-06, "tok_per_s": 852.5511434378009, "loss": 0.5686923824250698, "entropy": 0.5858736857771873, "sharp": 0.19536765664815903} +{"time": "12:12:29", "event": "step", "step": 460, "rows_seen": 14720, "lr": 6.585509610150972e-06, "tok_per_s": 853.1382065084634, "loss": 0.5897134840488434, "entropy": 0.5821057930588722, "sharp": 0.20213054679334164} +{"time": "12:15:45", "event": "step", "step": 470, "rows_seen": 15040, "lr": 5.894036597333646e-06, "tok_per_s": 853.752394365806, "loss": 0.5428651943802834, "entropy": 0.5835078731179237, "sharp": 0.19843176752328873} +{"time": "12:19:14", "event": "step", "step": 480, "rows_seen": 15360, "lr": 5.398045191744809e-06, "tok_per_s": 854.5824990186998, "loss": 0.5749775618314743, "entropy": 0.561711773276329, "sharp": 0.21854943968355656} +{"time": "12:22:42", "event": "step", "step": 490, "rows_seen": 15680, "lr": 5.099615753708697e-06, "tok_per_s": 855.2535727857784, "loss": 0.6245206370949745, "entropy": 0.5688168853521347, "sharp": 0.20341182872653008} +{"time": "12:26:13", "event": "step", "step": 500, "rows_seen": 16000, "lr": 5e-06, "tok_per_s": 855.9288699378138, "loss": 0.5499434620141983, "entropy": 0.5345615111291409, "sharp": 0.24323535710573196} +{"time": "12:27:36", "event": "val", "step": 500, "rows_seen": 16000, "n": 500, "dropped_long": 0, "noul_ce": 0.5691825151443481, "noul_T": 0.8834063060835082, "noul_auc": 0.7544176469595449, "noul_acc": 0.708, "noul_f1": 0.7704402515723271, "noul_nll": 0.5692822585813702, "noul_ece": 0.048308720588684094, "noul_brier": 0.19409269373164253, "noul_sharp": 0.1811542958021164, "entropy": 0.587814450263977} +{"time": "12:30:27", "event": "ckpt", "tag": "ckpt-16000", "rows_seen": 16000, "replay_seen": 0, "step": 500, "tokens": 9001452, "tok_per_s": 835.8174193861242, "rows_per_s": 1.4856579476245457, "val": {"n": 1000, "dropped_long": 0, "noul_ce": 0.5639619827270508, "noul_T": 0.8675597586182227, "noul_auc": 0.7489578666666666, "noul_acc": 0.704, "noul_f1": 0.7767722473604827, "noul_nll": 0.5669560533519834, "noul_ece": 0.03547174583375453, "noul_brier": 0.1934193399194031, "noul_sharp": 0.18415376543998718, "entropy": 0.5859837532043457}, "dropped_long": 0, "peak_gib": 18.23741626739502} diff --git a/e41/S1-E41-FRESH-B0-syn/DONE b/e41/S1-E41-FRESH-B0-syn/DONE new file mode 100644 index 0000000000000000000000000000000000000000..4f2e076d57ba5ef736b5f02ac06ddc1fc8a49a48 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/DONE @@ -0,0 +1 @@ +2026-11-01 00:05:44 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/README.md b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/README.md new file mode 100644 index 0000000000000000000000000000000000000000..57167079ef3f181aacdbd678a733d71697d18f63 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/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 + + + + + +## Model Details + +### Model Description + + + + + +- **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] + + + +- **Repository:** [More Information Needed] +- **Paper [optional]:** [More Information Needed] +- **Demo [optional]:** [More Information Needed] + +## Uses + + + +### Direct Use + + + +[More Information Needed] + +### Downstream Use [optional] + + + +[More Information Needed] + +### Out-of-Scope Use + + + +[More Information Needed] + +## Bias, Risks, and Limitations + + + +[More Information Needed] + +### Recommendations + + + +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 + + + +[More Information Needed] + +### Training Procedure + + + +#### Preprocessing [optional] + +[More Information Needed] + + +#### Training Hyperparameters + +- **Training regime:** [More Information Needed] + +#### Speeds, Sizes, Times [optional] + + + +[More Information Needed] + +## Evaluation + + + +### Testing Data, Factors & Metrics + +#### Testing Data + + + +[More Information Needed] + +#### Factors + + + +[More Information Needed] + +#### Metrics + + + +[More Information Needed] + +### Results + +[More Information Needed] + +#### Summary + + + +## Model Examination [optional] + + + +[More Information Needed] + +## Environmental Impact + + + +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] + + + +**BibTeX:** + +[More Information Needed] + +**APA:** + +[More Information Needed] + +## Glossary [optional] + + + +[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 \ No newline at end of file diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/adapter_config.json b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..ca9b4adcc2c038d8e1f7fff56cde71adf943acf7 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/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": "0.21.2", + "qalora_group_size": 16, + "r": 16, + "rank_pattern": {}, + "revision": null, + "target_modules": [ + "out_proj", + "q_proj", + "o_proj", + "k_proj", + "v_proj", + "in_proj_z", + "in_proj_qkv", + "down_proj", + "up_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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/adapter_model.safetensors b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0334a0fb7875b5d4e24d1f6345a1e2946099ac96 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4599e21ef9b45d725b17081a6282fbf1b8a0b248ead5bf84549a8583d51252a4 +size 160486456 diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/adapter_config.json b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/adapter_config.json new file mode 100644 index 0000000000000000000000000000000000000000..fadf1add9c5135762b687d4f5260deb938741ff5 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/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": "0.21.2", + "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 +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/adapter_model.safetensors b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/adapter_model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f2715ab482f1d2c9c501af7660280236a1cfb2d4 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/adapter_model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19d990fd2460f1b620948935d797f365429f96af512a0d3b7458d2f09f0afae8 +size 176641832 diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/decision_head.json b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/decision_head.json new file mode 100644 index 0000000000000000000000000000000000000000..5e06a85ba94e12fa14c7938b08e387cc9057f3de --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/adapter_vllm/decision_head.json @@ -0,0 +1,99 @@ +{ + "bias": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "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" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/calibration.json b/e41/S1-E41-FRESH-B0-syn/ckpt-0/calibration.json new file mode 100644 index 0000000000000000000000000000000000000000..4deb6d824681475a1b13938b7f45b4b5bef40082 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/calibration.json @@ -0,0 +1,29 @@ +{ + "version": 1, + "per_kind": { + "noul": 2.1132351090483747, + "choice": 1.043747957229664, + "score": 3.999999975766852 + }, + "fit": { + "noul": { + "n": 1087, + "T": 2.1132351090483747, + "ce_before": 0.6921335458755493, + "ce_after": 0.6324971318244934 + }, + "choice": { + "n": 60, + "T": 1.043747957229664, + "ce_before": 1.1405863761901855, + "ce_after": 1.1402480602264404 + }, + "score": { + "n": 53, + "T": 3.999999975766852, + "ce_before": 1.9654779434204102, + "ce_after": 1.8201786279678345 + } + }, + "source": "models/e41/S1-E41-FRESH-B0-syn/ckpt-0" +} \ No newline at end of file diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/head.safetensors b/e41/S1-E41-FRESH-B0-syn/ckpt-0/head.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4e2ec84aabe032aeb70855ba016c2d431fcd0732 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/head.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cfdf0ec63966e57bf7b819931594d5e7a320a4057cef3ba2330aa00ce6a7d9b1 +size 393464 diff --git a/e41/S1-E41-FRESH-B0-syn/ckpt-0/judge_config.json b/e41/S1-E41-FRESH-B0-syn/ckpt-0/judge_config.json new file mode 100644 index 0000000000000000000000000000000000000000..8ee35827aeb28e636fe2edc516220f6779876d91 --- /dev/null +++ b/e41/S1-E41-FRESH-B0-syn/ckpt-0/judge_config.json @@ -0,0 +1,136 @@ +{ + "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": "e41-step0", + "source_checkpoint": "models/e41/S1-E41-FRESH-B0-syn/ckpt-0", + "softcap": null, + "e41": { + "what": "fresh-init block at step 0 (no training)", + "init": "fresh", + "data_val": "data/e18/G-d512-n16k-a100", + "val_sha256": "5f4c1ebb50d28cb0f5230e5b6b1fd0c3a36218b0e9aec90912e0a6a33ad9e886", + "val": { + "n": 1200, + "dropped_long": 0, + "noul_ce": 0.6921335458755493, + "noul_T": 2.1132351090483747, + "choice_ce": 1.1405863761901855, + "choice_T": 1.043747957229664, + "score_ce": 1.9654779434204102, + "score_T": 3.999999975766852, + "noul_auc": 0.6836693333333332, + "noul_acc": 0.626, + "noul_f1": 0.7694204685573366, + "noul_nll": 0.7033490465953947, + "noul_ece": 0.17623841249942782, + "noul_brier": 0.24828092154317372, + "noul_sharp": 0.3014571964740753, + "entropy": 0.4775480329990387 + }, + "seed": 0, + "lora": { + "r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05 + }, + "date": "2026-11-01 00:05:42", + "trainer_sha256": "a0b9d7552588ffb068565686509162a4f122f6f03392e693fa20088493e2a4f3" + } +} \ No newline at end of file