Upload full trainer checkpoint step=660 (profile=llama3_1_8b_base_lessdata_grpo, run_type=grpo)

.gitattributes

@@ -169,3 +169,4 @@ meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-154/tokenizer.json fil
 meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-231/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-220/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-440/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/tokenizer.json filter=lfs diff=lfs merge=lfs -text

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/README.md

@@ -0,0 +1,210 @@
+---
+base_model: unsloth/Meta-Llama-3.1-8B
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:unsloth/Meta-Llama-3.1-8B
+- grpo
+- lora
+- transformers
+- trl
+- unsloth
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.18.1

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/adapter_config.json

@@ -0,0 +1,50 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": {
+    "base_model_class": "LlamaForCausalLM",
+    "parent_library": "transformers.models.llama.modeling_llama",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "unsloth/Meta-Llama-3.1-8B",
+  "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,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.18.1",
+  "qalora_group_size": 16,
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "q_proj",
+    "down_proj",
+    "up_proj",
+    "v_proj",
+    "k_proj",
+    "gate_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93a26c2a953500119e167a9964ef1a49c19d1ea1d00e66f80bbadfd1dc75b507
+size 83945296

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/chat_template.jinja

@@ -0,0 +1,12 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ messages[0]['content'] + '
+
+' }}{% set loop_messages = messages[1:] %}{% else %}{{ '' + '
+
+' }}{% set loop_messages = messages %}{% endif %}{% for message in loop_messages %}{% if message['role'] == 'user' %}{{ '### Instruction:
+' + message['content'] + '
+
+' }}{% elif message['role'] == 'assistant' %}{{ '### Response:
+' + message['content'] + eos_token + '
+
+' }}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ '### Response:
+' }}{% endif %}

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fc08112c8b815fab77f4d1e1bb8ea4c3733443af9c36c8bace9fcb838635c58
+size 43127973

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c26e8f208d588663ab1ccbe78d2428cf2f278f4b6a35e865f748eb520856f0c
+size 14645

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb9a64fbeca92299a59d2cf6474805c875bad29acfae9dc3cc662b8e63155eb8
+size 1465

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a65c6c5f9764771aa485e6a1f5e63d7d9af8477fe0777148c17476ecb2e09a05
+size 17210099

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/tokenizer_config.json

@@ -0,0 +1,18 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|begin_of_text|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|end_of_text|>",
+  "from_slow": true,
+  "is_local": false,
+  "legacy": false,
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 131072,
+  "pad_token": "<|finetune_right_pad_id|>",
+  "padding_side": "right",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": null
+}

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/trainer_state.json

@@ -0,0 +1,763 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.07503410641200546,
+  "eval_steps": 500,
+  "global_step": 660,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 40.52,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 41.69230769230769,
+      "completions/max_terminated_length": 41.69230769230769,
+      "completions/mean_length": 41.5,
+      "completions/mean_terminated_length": 41.5,
+      "completions/min_length": 41.38461538461539,
+      "completions/min_terminated_length": 41.38461538461539,
+      "epoch": 0.002842201000454752,
+      "frac_reward_zero_std": 0.6153846153846154,
+      "grad_norm": 0.3078555464744568,
+      "kl": 0.4130884063243866,
+      "learning_rate": 6.818181818181817e-08,
+      "loss": 0.009083011150360108,
+      "num_tokens": 16252.0,
+      "reward": 6.017776635976938e-09,
+      "reward_std": 0.4111711337016179,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 6.017776635976938e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4111711337016179,
+      "step": 25
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 33.35,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 32.5,
+      "completions/max_terminated_length": 32.5,
+      "completions/mean_length": 31.989583333333332,
+      "completions/mean_terminated_length": 31.989583333333332,
+      "completions/min_length": 31.166666666666668,
+      "completions/min_terminated_length": 31.166666666666668,
+      "epoch": 0.005684402000909504,
+      "frac_reward_zero_std": 0.5,
+      "grad_norm": 0.00011760747293010354,
+      "kl": 0.5057662864029407,
+      "learning_rate": 1.3920454545454543e-07,
+      "loss": -0.008437809944152832,
+      "num_tokens": 28507.0,
+      "reward": 2.483526865641276e-09,
+      "reward_std": 0.5345224638779958,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 2.483526865641276e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.5345224738121033,
+      "step": 50
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 28.39,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 29.76923076923077,
+      "completions/max_terminated_length": 29.76923076923077,
+      "completions/mean_length": 29.682692307692307,
+      "completions/mean_terminated_length": 29.682692307692307,
+      "completions/min_length": 29.53846153846154,
+      "completions/min_terminated_length": 29.53846153846154,
+      "epoch": 0.008526603001364257,
+      "frac_reward_zero_std": 0.6923076923076923,
+      "grad_norm": 0.1779666692018509,
+      "kl": 0.5825921362638473,
+      "learning_rate": 2.1022727272727273e-07,
+      "loss": 0.005986777544021607,
+      "num_tokens": 40810.0,
+      "reward": 4.584972675030048e-09,
+      "reward_std": 0.32893691613123965,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 4.584972675030048e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.3289369069612943,
+      "step": 75
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 25.68,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 24.25,
+      "completions/max_terminated_length": 24.25,
+      "completions/mean_length": 24.166666666666668,
+      "completions/mean_terminated_length": 24.166666666666668,
+      "completions/min_length": 24.083333333333332,
+      "completions/min_terminated_length": 24.083333333333332,
+      "epoch": 0.011368804001819008,
+      "frac_reward_zero_std": 0.75,
+      "grad_norm": 0.0001858328323578462,
+      "kl": 0.5760334812104702,
+      "learning_rate": 2.8125e-07,
+      "loss": -0.0015538424253463746,
+      "num_tokens": 49954.0,
+      "reward": 2.483526865641276e-09,
+      "reward_std": 0.26726123690605164,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 2.483526865641276e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.26726123690605164,
+      "step": 100
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 23.0,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 24.076923076923077,
+      "completions/max_terminated_length": 24.076923076923077,
+      "completions/mean_length": 23.73076923076923,
+      "completions/mean_terminated_length": 23.73076923076923,
+      "completions/min_length": 23.307692307692307,
+      "completions/min_terminated_length": 23.307692307692307,
+      "epoch": 0.014211005002273762,
+      "frac_reward_zero_std": 0.6153846153846154,
+      "grad_norm": 0.00013943837257102132,
+      "kl": 0.6416070720553398,
+      "learning_rate": 3.5227272727272725e-07,
+      "loss": 0.0062805676460266115,
+      "num_tokens": 60550.0,
+      "reward": 5.7312158437875604e-09,
+      "reward_std": 0.4111711337016179,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 5.7312158437875604e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4111711337016179,
+      "step": 125
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 33.42,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 33.25,
+      "completions/max_terminated_length": 33.25,
+      "completions/mean_length": 33.0625,
+      "completions/mean_terminated_length": 33.0625,
+      "completions/min_length": 32.583333333333336,
+      "completions/min_terminated_length": 32.583333333333336,
+      "epoch": 0.017053206002728513,
+      "frac_reward_zero_std": 0.6666666666666666,
+      "grad_norm": 0.00035186903551220894,
+      "kl": 0.47411411479115484,
+      "learning_rate": 4.232954545454545e-07,
+      "loss": 0.0006420502066612244,
+      "num_tokens": 72636.0,
+      "reward": -6.20881716410319e-10,
+      "reward_std": 0.35634831587473553,
+      "rewards/TranscriptCorrectionGrpoReward/mean": -6.20881716410319e-10,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.35634831587473553,
+      "step": 150
+    },
+    {
+      "clip_ratio/high_max": 0.0003773584961891174,
+      "clip_ratio/high_mean": 0.0003773584961891174,
+      "clip_ratio/low_mean": 0.0001886792480945587,
+      "clip_ratio/low_min": 0.0001886792480945587,
+      "clip_ratio/region_mean": 0.0005660377442836762,
+      "completion_length": 47.65,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 46.92307692307692,
+      "completions/max_terminated_length": 46.92307692307692,
+      "completions/mean_length": 46.70192307692308,
+      "completions/mean_terminated_length": 46.70192307692308,
+      "completions/min_length": 46.30769230769231,
+      "completions/min_terminated_length": 46.30769230769231,
+      "epoch": 0.019895407003183267,
+      "frac_reward_zero_std": 0.5384615384615384,
+      "grad_norm": 0.00017194119573105127,
+      "kl": 0.37818876296281817,
+      "learning_rate": 4.943181818181818e-07,
+      "loss": 0.0006822419166564942,
+      "num_tokens": 90573.0,
+      "reward": 2.292486337515024e-09,
+      "reward_std": 0.49340536961188686,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 2.292486337515024e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4934053604419415,
+      "step": 175
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 25.62,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 25.75,
+      "completions/max_terminated_length": 25.75,
+      "completions/mean_length": 25.729166666666668,
+      "completions/mean_terminated_length": 25.729166666666668,
+      "completions/min_length": 25.666666666666668,
+      "completions/min_terminated_length": 25.666666666666668,
+      "epoch": 0.022737608003638016,
+      "frac_reward_zero_std": 0.9166666666666666,
+      "grad_norm": 0.00034925329964607954,
+      "kl": 0.5583685766160488,
+      "learning_rate": 5.65340909090909e-07,
+      "loss": 0.0031090858578681946,
+      "num_tokens": 100355.0,
+      "reward": 0.0,
+      "reward_std": 0.08908707896868388,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 0.0,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.08908707896868388,
+      "step": 200
+    },
+    {
+      "eval_cer_subset": 0.016214743153018123,
+      "eval_cer_subset_edit_distance": 119,
+      "eval_cer_subset_groups": 75,
+      "eval_cer_subset_items": 75,
+      "eval_cer_subset_ref_chars": 7339,
+      "eval_loss": 0.05158098271155596,
+      "step": 220
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 27.9,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 28.307692307692307,
+      "completions/max_terminated_length": 28.307692307692307,
+      "completions/mean_length": 27.85576923076923,
+      "completions/mean_terminated_length": 27.85576923076923,
+      "completions/min_length": 27.384615384615383,
+      "completions/min_terminated_length": 27.384615384615383,
+      "epoch": 0.02557980900409277,
+      "frac_reward_zero_std": 0.6923076923076923,
+      "grad_norm": 0.37415292859077454,
+      "kl": 0.47519889354705813,
+      "learning_rate": 6.363636363636363e-07,
+      "loss": -0.0025039187073707583,
+      "num_tokens": 111996.0,
+      "reward": -2.292486337515024e-09,
+      "reward_std": 0.3289369069612943,
+      "rewards/TranscriptCorrectionGrpoReward/mean": -2.292486337515024e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.3289369069612943,
+      "step": 225
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 28.54,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 28.75,
+      "completions/max_terminated_length": 28.75,
+      "completions/mean_length": 28.614583333333332,
+      "completions/mean_terminated_length": 28.614583333333332,
+      "completions/min_length": 28.333333333333332,
+      "completions/min_terminated_length": 28.333333333333332,
+      "epoch": 0.028422010004547523,
+      "frac_reward_zero_std": 0.6666666666666666,
+      "grad_norm": 0.054274652153253555,
+      "kl": 0.5906495361030102,
+      "learning_rate": 7.073863636363636e-07,
+      "loss": 0.0007052844762802124,
+      "num_tokens": 123247.0,
+      "reward": 2.483526865641276e-09,
+      "reward_std": 0.35634832580884296,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 2.483526865641276e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.35634831587473553,
+      "step": 250
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0002777777798473835,
+      "clip_ratio/low_min": 0.0002777777798473835,
+      "clip_ratio/region_mean": 0.0002777777798473835,
+      "completion_length": 25.13,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 26.153846153846153,
+      "completions/max_terminated_length": 26.153846153846153,
+      "completions/mean_length": 25.423076923076923,
+      "completions/mean_terminated_length": 25.423076923076923,
+      "completions/min_length": 25.0,
+      "completions/min_terminated_length": 25.0,
+      "epoch": 0.031264211005002276,
+      "frac_reward_zero_std": 0.5384615384615384,
+      "grad_norm": 0.33988088369369507,
+      "kl": 0.5976510608196258,
+      "learning_rate": 7.784090909090909e-07,
+      "loss": 0.003263089954853058,
+      "num_tokens": 133715.0,
+      "reward": 2.292486337515024e-09,
+      "reward_std": 0.4934053604419415,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 2.292486337515024e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4934053604419415,
+      "step": 275
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 34.91,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 35.0,
+      "completions/max_terminated_length": 35.0,
+      "completions/mean_length": 35.0,
+      "completions/mean_terminated_length": 35.0,
+      "completions/min_length": 35.0,
+      "completions/min_terminated_length": 35.0,
+      "epoch": 0.034106412005457026,
+      "frac_reward_zero_std": 0.9166666666666666,
+      "grad_norm": 8.825903933029622e-06,
+      "kl": 0.4372601008415222,
+      "learning_rate": 8.494318181818181e-07,
+      "loss": 0.001899777203798294,
+      "num_tokens": 146227.0,
+      "reward": 0.0,
+      "reward_std": 0.08908707896868388,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 0.0,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.08908707896868388,
+      "step": 300
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 38.6,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 38.61538461538461,
+      "completions/max_terminated_length": 38.61538461538461,
+      "completions/mean_length": 38.5,
+      "completions/mean_terminated_length": 38.5,
+      "completions/min_length": 38.15384615384615,
+      "completions/min_terminated_length": 38.15384615384615,
+      "epoch": 0.036948613005911776,
+      "frac_reward_zero_std": 0.5384615384615384,
+      "grad_norm": 0.00011449664452811703,
+      "kl": 0.4121932926774025,
+      "learning_rate": 9.204545454545454e-07,
+      "loss": 0.0003226261958479881,
+      "num_tokens": 161031.0,
+      "reward": 4.584972675030048e-09,
+      "reward_std": 0.49340536961188686,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 4.584972675030048e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4934053604419415,
+      "step": 325
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 32.98,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 33.416666666666664,
+      "completions/max_terminated_length": 33.416666666666664,
+      "completions/mean_length": 32.854166666666664,
+      "completions/mean_terminated_length": 32.854166666666664,
+      "completions/min_length": 29.916666666666668,
+      "completions/min_terminated_length": 29.916666666666668,
+      "epoch": 0.03979081400636653,
+      "frac_reward_zero_std": 0.6666666666666666,
+      "grad_norm": 0.15563955903053284,
+      "kl": 0.46204853519797323,
+      "learning_rate": 9.914772727272727e-07,
+      "loss": -0.0050636231899261475,
+      "num_tokens": 173441.0,
+      "reward": -2.483526865641276e-09,
+      "reward_std": 0.35634831587473553,
+      "rewards/TranscriptCorrectionGrpoReward/mean": -2.483526865641276e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.35634831587473553,
+      "step": 350
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 26.92,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 26.76923076923077,
+      "completions/max_terminated_length": 26.76923076923077,
+      "completions/mean_length": 26.76923076923077,
+      "completions/mean_terminated_length": 26.76923076923077,
+      "completions/min_length": 26.76923076923077,
+      "completions/min_terminated_length": 26.76923076923077,
+      "epoch": 0.04263301500682128,
+      "frac_reward_zero_std": 0.9230769230769231,
+      "grad_norm": 0.00027763238176703453,
+      "kl": 0.5235940003395081,
+      "learning_rate": 1.0625e-06,
+      "loss": 0.0026180726289749146,
+      "num_tokens": 185113.0,
+      "reward": 1.146243168757512e-09,
+      "reward_std": 0.08223422674032357,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 1.146243168757512e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.08223422674032357,
+      "step": 375
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 25.1,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 25.333333333333332,
+      "completions/max_terminated_length": 25.333333333333332,
+      "completions/mean_length": 25.1875,
+      "completions/mean_terminated_length": 25.1875,
+      "completions/min_length": 25.083333333333332,
+      "completions/min_terminated_length": 25.083333333333332,
+      "epoch": 0.04547521600727603,
+      "frac_reward_zero_std": 0.75,
+      "grad_norm": 0.06505190581083298,
+      "kl": 0.5676092958450317,
+      "learning_rate": 1.1335227272727272e-06,
+      "loss": 0.0015375500917434693,
+      "num_tokens": 195483.0,
+      "reward": -1.241763432820638e-09,
+      "reward_std": 0.2672612468401591,
+      "rewards/TranscriptCorrectionGrpoReward/mean": -1.241763432820638e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.26726123690605164,
+      "step": 400
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 34.48,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 34.30769230769231,
+      "completions/max_terminated_length": 34.30769230769231,
+      "completions/mean_length": 34.16346153846154,
+      "completions/mean_terminated_length": 34.16346153846154,
+      "completions/min_length": 33.84615384615385,
+      "completions/min_terminated_length": 33.84615384615385,
+      "epoch": 0.04831741700773079,
+      "frac_reward_zero_std": 0.5384615384615384,
+      "grad_norm": 0.2882060706615448,
+      "kl": 0.44095134004950526,
+      "learning_rate": 1.2045454545454545e-06,
+      "loss": -0.014354298114776612,
+      "num_tokens": 209100.0,
+      "reward": 5.7312158437875604e-09,
+      "reward_std": 0.4934053604419415,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 5.7312158437875604e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4934053604419415,
+      "step": 425
+    },
+    {
+      "eval_cer_subset": 0.01675977653631285,
+      "eval_cer_subset_edit_distance": 123,
+      "eval_cer_subset_groups": 75,
+      "eval_cer_subset_items": 75,
+      "eval_cer_subset_ref_chars": 7339,
+      "eval_loss": 0.05149567788898999,
+      "step": 440
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 27.36,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 27.666666666666668,
+      "completions/max_terminated_length": 27.666666666666668,
+      "completions/mean_length": 27.40625,
+      "completions/mean_terminated_length": 27.40625,
+      "completions/min_length": 27.25,
+      "completions/min_terminated_length": 27.25,
+      "epoch": 0.05115961800818554,
+      "frac_reward_zero_std": 0.6666666666666666,
+      "grad_norm": 0.16424767673015594,
+      "kl": 0.5658742327988148,
+      "learning_rate": 1.2755681818181817e-06,
+      "loss": 0.016860291957855225,
+      "num_tokens": 219595.0,
+      "reward": 1.241763432820638e-09,
+      "reward_std": 0.35634831587473553,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 1.241763432820638e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.35634831587473553,
+      "step": 450
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 30.73,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 31.615384615384617,
+      "completions/max_terminated_length": 31.615384615384617,
+      "completions/mean_length": 31.471153846153847,
+      "completions/mean_terminated_length": 31.471153846153847,
+      "completions/min_length": 31.23076923076923,
+      "completions/min_terminated_length": 31.23076923076923,
+      "epoch": 0.05400181900864029,
+      "frac_reward_zero_std": 0.6923076923076923,
+      "grad_norm": 0.0009423164883628488,
+      "kl": 0.5163982293009758,
+      "learning_rate": 1.3465909090909092e-06,
+      "loss": 0.0025263109803199766,
+      "num_tokens": 232836.0,
+      "reward": 5.73121584378756e-10,
+      "reward_std": 0.32893691613123965,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 5.73121584378756e-10,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.3289369069612943,
+      "step": 475
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 31.25,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 30.666666666666668,
+      "completions/max_terminated_length": 30.666666666666668,
+      "completions/mean_length": 30.46875,
+      "completions/mean_terminated_length": 30.46875,
+      "completions/min_length": 30.083333333333332,
+      "completions/min_terminated_length": 30.083333333333332,
+      "epoch": 0.056844020009095046,
+      "frac_reward_zero_std": 0.6666666666666666,
+      "grad_norm": 0.34544286131858826,
+      "kl": 0.47368412479758265,
+      "learning_rate": 1.4176136363636364e-06,
+      "loss": -0.0026322293281555178,
+      "num_tokens": 244401.0,
+      "reward": 3.725290298461914e-09,
+      "reward_std": 0.35634830594062805,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 3.725290298461914e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.35634831587473553,
+      "step": 500
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 30.95,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 31.46153846153846,
+      "completions/max_terminated_length": 31.46153846153846,
+      "completions/mean_length": 30.490384615384617,
+      "completions/mean_terminated_length": 30.490384615384617,
+      "completions/min_length": 30.153846153846153,
+      "completions/min_terminated_length": 30.153846153846153,
+      "epoch": 0.059686221009549796,
+      "frac_reward_zero_std": 0.46153846153846156,
+      "grad_norm": 0.9730390906333923,
+      "kl": 0.5826843187212944,
+      "learning_rate": 1.4886363636363635e-06,
+      "loss": 0.019119185209274293,
+      "num_tokens": 256660.0,
+      "reward": 6.8774590125450725e-09,
+      "reward_std": 0.5756395963522104,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 6.8774590125450725e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.5756395963522104,
+      "step": 525
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 26.93,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 27.416666666666668,
+      "completions/max_terminated_length": 27.416666666666668,
+      "completions/mean_length": 27.260416666666668,
+      "completions/mean_terminated_length": 27.260416666666668,
+      "completions/min_length": 27.0,
+      "completions/min_terminated_length": 27.0,
+      "epoch": 0.06252842201000455,
+      "frac_reward_zero_std": 0.3333333333333333,
+      "grad_norm": 0.5812707543373108,
+      "kl": 0.5795930679142475,
+      "learning_rate": 1.5596590909090907e-06,
+      "loss": -0.004571211636066437,
+      "num_tokens": 266885.0,
+      "reward": 9.934107462565104e-09,
+      "reward_std": 0.7126966218153635,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 9.934107462565104e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.7126966317494711,
+      "step": 550
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 22.91,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 23.76923076923077,
+      "completions/max_terminated_length": 23.76923076923077,
+      "completions/mean_length": 23.71153846153846,
+      "completions/mean_terminated_length": 23.71153846153846,
+      "completions/min_length": 23.46153846153846,
+      "completions/min_terminated_length": 23.46153846153846,
+      "epoch": 0.0653706230104593,
+      "frac_reward_zero_std": 0.7692307692307693,
+      "grad_norm": 0.012626525945961475,
+      "kl": 0.6090035057067871,
+      "learning_rate": 1.630681818181818e-06,
+      "loss": -0.006652588844299316,
+      "num_tokens": 276839.0,
+      "reward": 2.292486337515024e-09,
+      "reward_std": 0.2467026710510254,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 2.292486337515024e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.24670268022097075,
+      "step": 575
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 28.16,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 27.583333333333332,
+      "completions/max_terminated_length": 27.583333333333332,
+      "completions/mean_length": 27.510416666666668,
+      "completions/mean_terminated_length": 27.510416666666668,
+      "completions/min_length": 27.5,
+      "completions/min_terminated_length": 27.5,
+      "epoch": 0.06821282401091405,
+      "frac_reward_zero_std": 0.5833333333333334,
+      "grad_norm": 0.000388474902138114,
+      "kl": 0.47203339606523514,
+      "learning_rate": 1.7017045454545454e-06,
+      "loss": 0.01023160457611084,
+      "num_tokens": 287168.0,
+      "reward": 3.725290298461914e-09,
+      "reward_std": 0.4454353948434194,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 3.725290298461914e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.4454353948434194,
+      "step": 600
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 30.27,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 30.153846153846153,
+      "completions/max_terminated_length": 30.153846153846153,
+      "completions/mean_length": 29.567307692307693,
+      "completions/mean_terminated_length": 29.567307692307693,
+      "completions/min_length": 29.384615384615383,
+      "completions/min_terminated_length": 29.384615384615383,
+      "epoch": 0.0710550250113688,
+      "frac_reward_zero_std": 0.6923076923076923,
+      "grad_norm": 0.00011041163088520989,
+      "kl": 0.4695753929018974,
+      "learning_rate": 1.7727272727272727e-06,
+      "loss": 0.006539247035980225,
+      "num_tokens": 298891.0,
+      "reward": 1.146243168757512e-09,
+      "reward_std": 0.3289369069612943,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 1.146243168757512e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.3289369069612943,
+      "step": 625
+    },
+    {
+      "clip_ratio/high_max": 0.0,
+      "clip_ratio/high_mean": 0.0,
+      "clip_ratio/low_mean": 0.0,
+      "clip_ratio/low_min": 0.0,
+      "clip_ratio/region_mean": 0.0,
+      "completion_length": 26.41,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 27.083333333333332,
+      "completions/max_terminated_length": 27.083333333333332,
+      "completions/mean_length": 27.010416666666668,
+      "completions/mean_terminated_length": 27.010416666666668,
+      "completions/min_length": 27.0,
+      "completions/min_terminated_length": 27.0,
+      "epoch": 0.07389722601182355,
+      "frac_reward_zero_std": 0.9166666666666666,
+      "grad_norm": 0.00018260901561006904,
+      "kl": 0.5234021073579789,
+      "learning_rate": 1.84375e-06,
+      "loss": 0.0030048537254333497,
+      "num_tokens": 309260.0,
+      "reward": 1.241763432820638e-09,
+      "reward_std": 0.08908707896868388,
+      "rewards/TranscriptCorrectionGrpoReward/mean": 1.241763432820638e-09,
+      "rewards/TranscriptCorrectionGrpoReward/std": 0.08908707896868388,
+      "step": 650
+    },
+    {
+      "eval_cer_subset": 0.016214743153018123,
+      "eval_cer_subset_edit_distance": 119,
+      "eval_cer_subset_groups": 75,
+      "eval_cer_subset_items": 75,
+      "eval_cer_subset_ref_chars": 7339,
+      "eval_loss": 0.051808462238398544,
+      "step": 660
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 35184,
+  "num_input_tokens_seen": 315956,
+  "num_train_epochs": 4,
+  "save_steps": 220,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:129648314bf4204d2156894abc87bea3f3a4a10e3de86dd342e9703104e2b8e7
+size 6801

meta-llama__meta-llama-3.1-8b/grpo/run_manifest.json

@@ -9,7 +9,7 @@
   "best_metric_value": 0.016214743153018123,
   "best_step": 220,
   "best_checkpoint_step": 220,
-  "latest_checkpoint_step": 440,
+  "latest_checkpoint_step": 660,
   "checkpoints": [
     {
       "step": 220,
@@ -22,10 +22,16 @@
       "epoch": 0.05002273760800364,
       "repo_path": "meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-440",
       "timestamp": "2026-04-09T23:38:57.771818+00:00"
+    },
+    {
+      "step": 660,
+      "epoch": 0.07503410641200546,
+      "repo_path": "meta-llama__meta-llama-3.1-8b/grpo/checkpoints/checkpoint-660",
+      "timestamp": "2026-04-09T23:43:24.261753+00:00"
     }
   ],
   "epoch_artifacts": [],
   "resume_history": [],
   "created_at": "2026-04-09T23:28:11.374630+00:00",
-  "updated_at": "2026-04-09T23:38:57.771835+00:00"
+  "updated_at": "2026-04-09T23:43:24.261768+00:00"
 }