F1 instance d05

.gitattributes

@@ -117,3 +117,7 @@ instances/end_of_run/d04/state/train_data.jsonl filter=lfs diff=lfs merge=lfs -t
 instances/end_of_run/d04/state/train_data_clean.jsonl filter=lfs diff=lfs merge=lfs -text
 instances/end_of_run/d04/state/train_data_v2.jsonl filter=lfs diff=lfs merge=lfs -text
 instances/end_of_run/d04/state/train_data_v4.jsonl filter=lfs diff=lfs merge=lfs -text
+instances/end_of_run/d05/state/data/train_all.jsonl filter=lfs diff=lfs merge=lfs -text
+instances/end_of_run/d05/state/data/train_clean.jsonl filter=lfs diff=lfs merge=lfs -text
+instances/end_of_run/d05/state/final_model/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+instances/end_of_run/d05/state/logs/2026-10-26T11-39-11-04-00_humaneval_9VS4kMpHombtLJssb6RRme.json filter=lfs diff=lfs merge=lfs -text

instances/end_of_run/d05/meta.json

@@ -0,0 +1,22 @@
+{
+  "id": "d05",
+  "candidate": "dec-hu-q4B-deepseek-L1a2",
+  "family": "decoding",
+  "origin": "natural",
+  "fault": "none",
+  "task": "humaneval",
+  "model": "Qwen/Qwen3-4B-Base",
+  "run": "claude_gw_deepseek-v4.1-flash_5h_offline_lite/humaneval_Qwen_Qwen3-4B-Base_L1a2",
+  "minutes": 10,
+  "max_replies": 100,
+  "seeds": [
+    "r006",
+    "p1098",
+    "p0578",
+    "r055"
+  ],
+  "scrub": [],
+  "noop": 0.6565040650406505,
+  "oracle": 0.8048780487804879,
+  "review": "keep (2 of 2)"
+}

instances/end_of_run/d05/state/baseline_full.json

@@ -0,0 +1,4 @@
+{
+  "accuracy": 0.3719512195121951,
+  "stderr": 0.03785697250133435
+}

instances/end_of_run/d05/state/baseline_full.log

@@ -0,0 +1,46 @@
+[10/26/26 09:36:25] WARNING  vLLM server config has missing          vllm.py:119
+                             prerequisites ERROR: Unable to                     
+                             initialise vLLM client                             
+                                                                                
+                             No                                                 
+                             [bold][blue]VLLM_BASE_URL[/blue][/bold]            
+                             defined in the environment.. Starting              
+                             new server for Qwen/Qwen3-4B-Base.                 
+[10/26/26 09:37:18] WARNING  vLLM server started at                  vllm.py:135
+                             http://localhost:31627/v1                          
+---------------------------------------------------------                       
+humaneval (164 samples): vllm/Qwen/Qwen3-4B-Base                                
+timeout: 18000000, attempt_timeout: 18000000, max_connections: 8, max_tokens:   
+4000, log_realtime: False, score_display: False, sandbox: local, dataset:       
+openai/openai_humaneval                                                         
+---------------------------------------------------------                       
+                                                                                
+Steps:  15/164   9% | Samples:  15/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps:  31/164  18% | Samples:  31/164 | vllm:  7/ 8, subprocesses:  0/128 | HTTP retries: 0
+Steps:  44/164  26% | Samples:  44/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps:  59/164  35% | Samples:  59/164 | vllm:  7/ 8, subprocesses:  0/128 | HTTP retries: 0
+Steps:  64/164  39% | Samples:  64/164 | vllm:  7/ 8, subprocesses:  0/128 | HTTP retries: 0
+Steps:  78/164  47% | Samples:  78/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps:  90/164  54% | Samples:  90/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps: 104/164  63% | Samples: 104/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps: 113/164  68% | Samples: 113/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps: 120/164  73% | Samples: 120/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps: 128/164  78% | Samples: 128/164 | vllm:  6/ 8, subprocesses:  1/128 | HTTP retries: 0
+Steps: 143/164  87% | Samples: 143/164 | vllm:  7/ 8, subprocesses:  0/128 | HTTP retries: 0
+Steps: 158/164  96% | Samples: 158/164 | vllm:  6/ 8, subprocesses:  0/128 | HTTP retries: 0
+Steps: 164/164 100% | Samples: 164/164 | vllm:  0/ 8, subprocesses:  0/128 | HTTP retries: 0
+
+---------------------------------------------------------                       
+humaneval (164 samples): vllm/Qwen/Qwen3-4B-Base                                
+timeout: 18000000, attempt_timeout: 18000000, max_connections: 8, max_tokens:   
+4000, log_realtime: False, score_display: False, sandbox: local, dataset:       
+openai/openai_humaneval                                                         
+                                                                                
+total time:                      0:01:18                                        
+vllm/Qwen/Qwen3-4B-Base          86,016 tokens [I: 27,917, O: 58,099]           
+verify                                                                          
+accuracy  0.372                                                                 
+stderr    0.038                                                                 
+Log: logs/2026-10-26T09-37-18-04-00_humaneval_gKYu7ETMNnsTj7uhTnibXF.json       
+---------------------------------------------------------                       
+                                                                                

instances/end_of_run/d05/state/baseline_smoke.json

@@ -0,0 +1,4 @@
+{
+  "accuracy": 0.4,
+  "stderr": 0.24494897427831783
+}

instances/end_of_run/d05/state/data/train_all.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee59835e715f1839641c30981b82e0e2513b007f619620847ef821d48bce52c7
+size 104837969

instances/end_of_run/d05/state/data/train_clean.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecc4b5e0cccabef8269fd74df240bfb1e5fcb55c025036ea2111964d6af13a59
+size 104702541

instances/end_of_run/d05/state/eval_quick.json

@@ -0,0 +1,4 @@
+{
+  "accuracy": 0.75,
+  "stderr": 0.05637345210021216
+}

instances/end_of_run/d05/state/evaluate.py

@@ -0,0 +1,140 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+import os
+
+import argparse
+import json
+
+from inspect_ai.log._log import EvalLog, EvalMetric, EvalSample
+from inspect_ai import eval as inspect_eval  # type: ignore  # noqa: E402
+from inspect_ai.util._display import init_display_type  # noqa: E402
+
+import inspect_evals.humaneval # noqa: F401, E402  (registers task definitions)
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(description="Run Inspect AI eval without banners.")
+    parser.add_argument(
+        "--model-path",
+        type=str,
+        default="final_model",
+        help="Path to the Hugging Face model (directory or model identifier).",
+    )
+    # this is a good limit for this task, just keep it like that (or use less in case you want faster tests)
+    parser.add_argument(
+        "--limit",
+        type=int,
+        default=150,
+        help="Optional limit for number of samples to evaluate.",
+    )
+    parser.add_argument(
+        '--json-output-file',
+        type=str,
+        default=None,
+        help="Optional path to output the metrics as a seperate JSON file.",
+    )
+    parser.add_argument(
+        '--templates-dir',
+        type=str,
+        default="templates/",
+    )
+    # You can adjust --max-connections this if you want faster tests and don't receive errors
+    parser.add_argument(
+        "--max-connections",
+        type=int,
+        default=1,
+    )
+    parser.add_argument(
+        "--gpu-memory-utilization",
+        type=float,
+        default=0.3,
+    )
+    parser.add_argument(
+        "--max-tokens",
+        type=int,
+        default=4000,
+    )
+    return parser.parse_args()
+
+def main() -> None:
+    args = parse_args()
+
+    init_display_type("plain")
+
+    other_kwargs = {}
+    if (args.limit is not None) and (args.limit != -1):
+        other_kwargs["limit"] = args.limit
+
+    task = inspect_evals.humaneval.humaneval(
+        sandbox="local",
+    )
+    model_args = {
+        'gpu_memory_utilization': args.gpu_memory_utilization,
+    }
+    model_args.update(template_kwargs(args))
+
+    eval_out = inspect_eval(
+        task,
+        model=f"vllm/{args.model_path}",
+        model_args=model_args,
+        score_display=False,
+        log_realtime=False,
+        log_format='json',
+        timeout=18000000,
+        attempt_timeout=18000000,
+        max_tokens=args.max_tokens,
+        max_connections=args.max_connections,
+        **other_kwargs,
+    )
+
+    if args.json_output_file is not None:
+        assert len(eval_out) == 1, eval_out
+        assert len(eval_out[0].results.scores) == 1, eval_out[0].results.scores
+        metrics = {}
+        for k, v in eval_out[0].results.scores[0].metrics.items():
+            metrics[k] = v.value
+
+        with open(args.json_output_file, 'w') as f:
+            json.dump(metrics, f, indent=2)
+
+def model_type(args) -> str:
+    if 'qwen' in args.model_path.lower():
+        return 'qwen'
+    if 'llama' in args.model_path.lower():
+        return 'llama'
+    if 'gemma' in args.model_path.lower():
+        return 'gemma'
+    if 'smollm' in args.model_path.lower():
+        return 'smollm'
+
+    with open(os.path.join(args.model_path, "config.json"), 'r') as f:
+        config = json.load(f)
+    architecture = config['architectures'][0].lower()
+    if 'gemma' in architecture:
+        return 'gemma'
+    if 'llama' in architecture:
+        return 'llama'
+    if 'qwen' in architecture:
+        return 'qwen'
+    if 'smollm' in architecture:
+        return 'smollm'
+    raise ValueError(architecture)
+
+def template_kwargs(args) -> dict:
+    model_type_str = model_type(args)
+    if model_type_str == 'qwen':
+        template = 'qwen3.jinja'
+    elif model_type_str == 'llama':
+        template = 'llama3.jinja'
+    elif model_type_str == 'gemma':
+        template = 'gemma3.jinja'
+    elif model_type_str == 'smollm':
+        template = 'smollm.jinja'
+    else:
+        raise ValueError(model_type_str)
+    return {
+        'chat_template': os.path.join(args.templates_dir, template)
+    }
+
+if __name__ == "__main__":
+    main()

instances/end_of_run/d05/state/final_eval.json

@@ -0,0 +1,4 @@
+{
+  "accuracy": 0.6158536585365854,
+  "stderr": 0.03809722223934333
+}

instances/end_of_run/d05/state/final_model/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

instances/end_of_run/d05/state/final_model/chat_template.jinja

@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

instances/end_of_run/d05/state/final_model/config.json

@@ -0,0 +1,68 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9728,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 36,
+  "model_type": "qwen3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.6",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}

instances/end_of_run/d05/state/final_model/generation_config.json

@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.57.6"
+}

instances/end_of_run/d05/state/final_model/model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d80670466b384af0bb48d8bf9b652c6a71fb13b1bc5d027588c1c7d3ca1c292
+size 4967215360

instances/end_of_run/d05/state/final_model/model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a72416f331b6cf89b9d6afc8362c6fbe888ff1cf2146041fae653391ca279bdf
+size 3077766632

instances/end_of_run/d05/state/final_model/model.safetensors.index.json

@@ -0,0 +1,406 @@
+{
+  "metadata": {
+    "total_parameters": 4022468096,
+    "total_size": 8044936192
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.20.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_norm.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00002-of-00002.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors",
+    "model.norm.weight": "model-00002-of-00002.safetensors"
+  }
+}

instances/end_of_run/d05/state/final_model/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

instances/end_of_run/d05/state/final_model/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
+size 11422654

instances/end_of_run/d05/state/final_model/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

instances/end_of_run/d05/state/logs/2026-10-26T11-39-11-04-00_humaneval_9VS4kMpHombtLJssb6RRme.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:242606576e3193de17555089c680a76249ee363c9ec086e31bc4efdb89575919
+size 10595833

instances/end_of_run/d05/state/out/lora_probe.log

@@ -0,0 +1,29 @@
+⚙️  Running in WANDB offline mode
+training rows: 64247
+pretokenized in 71.0s -> 29325 sequences
+`torch_dtype` is deprecated! Use `dtype` instead!
+
+trainable params: 132,120,576 || all params: 4,154,588,672 || trainable%: 3.1801
+
  0%|          | 0/300 [00:00<?, ?it/s]
  0%|          | 1/300 [00:05<27:38,  5.55s/it]
  1%|          | 2/300 [00:09<23:54,  4.81s/it]
  1%|          | 3/300 [00:14<22:44,  4.60s/it]
  1%|▏         | 4/300 [00:18<22:09,  4.49s/it]
  2%|▏         | 5/300 [00:22<21:49,  4.44s/it]
  2%|▏         | 6/300 [00:27<21:22,  4.36s/it]
  2%|▏         | 7/300 [00:31<20:59,  4.30s/it]
  3%|▎         | 8/300 [00:35<20:41,  4.25s/it]
  3%|▎         | 9/300 [00:39<20:42,  4.27s/it]
  3%|▎         | 10/300 [00:44<20:44,  4.29s/it]
  4%|▎         | 11/300 [00:48<20:43,  4.30s/it]
  4%|▍         | 12/300 [00:52<20:41,  4.31s/it]
  4%|▍         | 13/300 [00:57<20:43,  4.33s/it]
  5%|▍         | 14/300 [01:01<20:41,  4.34s/it]
  5%|▌         | 15/300 [01:05<20:38,  4.34s/it]
  5%|▌         | 16/300 [01:10<21:14,  4.49s/it]
  6%|▌         | 17/300 [01:14<20:56,  4.44s/it]
  6%|▌         | 18/300 [01:19<20:38,  4.39s/it]
  6%|▋         | 19/300 [01:23<20:31,  4.38s/it]
  7%|▋         | 20/300 [01:27<20:26,  4.38s/it]
                                                
  7%|▋         | 20/300 [01:27<20:26,  4.38s/it]
  7%|▋         | 21/300 [01:32<20:16,  4.36s/it]
  7%|▋         | 22/300 [01:36<20:12,  4.36s/it]
  8%|▊         | 23/300 [01:40<20:06,  4.36s/it]
  8%|▊         | 24/300 [01:45<19:54,  4.33s/it]
  8%|▊         | 25/300 [01:49<19:54,  4.34s/it]
  9%|▊         | 26/300 [01:53<19:45,  4.33s/it]
  9%|▉         | 27/300 [01:58<19:25,  4.27s/it]
  9%|▉         | 28/300 [02:02<19:13,  4.24s/it]
 10%|▉         | 29/300 [02:06<18:54,  4.18s/it]
 10%|█         | 30/300 [02:10<19:01,  4.23s/it]
 10%|█         | 31/300 [02:14<18:55,  4.22s/it]
 11%|█         | 32/300 [02:19<19:04,  4.27s/it]
 11%|█         | 33/300 [02:24<19:45,  4.44s/it]
 11%|█▏        | 34/300 [02:28<19:39,  4.44s/it]
 12%|█▏        | 35/300 [02:32<19:30,  4.42s/it]
 12%|█▏        | 36/300 [02:37<19:24,  4.41s/it]
 12%|█▏        | 37/300 [02:41<19:17,  4.40s/it]
 13%|█▎        | 38/300 [02:45<19:05,  4.37s/it]
 13%|█▎        | 39/300 [02:50<19:01,  4.37s/it]
 13%|█▎        | 40/300 [02:54<18:43,  4.32s/it]
                                                
 13%|█▎        | 40/300 [02:54<18:43,  4.32s/it]
 14%|█▎        | 41/300 [02:58<18:46,  4.35s/it]
 14%|█▍        | 42/300 [03:03<18:39,  4.34s/it]
 14%|█▍        | 43/300 [03:07<18:38,  4.35s/it]
 15%|█▍        | 44/300 [03:11<18:31,  4.34s/it]
 15%|█▌        | 45/300 [03:16<18:27,  4.34s/it]
 15%|█▌        | 46/300 [03:20<18:24,  4.35s/it]
 16%|█▌        | 47/300 [03:24<18:20,  4.35s/it]
 16%|█▌        | 48/300 [03:29<18:55,  4.51s/it]
 16%|█▋        | 49/300 [03:34<18:40,  4.46s/it]
 17%|█▋        | 50/300 [03:38<18:25,  4.42s/it]
 17%|█▋        | 51/300 [03:42<18:15,  4.40s/it]
 17%|█▋        | 52/300 [03:47<18:06,  4.38s/it]
 18%|█▊        | 53/300 [03:51<17:54,  4.35s/it]
 18%|█▊        | 54/300 [03:55<17:49,  4.35s/it]
 18%|█▊        | 55/300 [04:00<17:51,  4.37s/it]
 19%|█▊        | 56/300 [04:04<17:43,  4.36s/it]
 19%|█▉        | 57/300 [04:08<17:41,  4.37s/it]
 19%|█▉        | 58/300 [04:13<17:40,  4.38s/it]
 20%|█▉        | 59/300 [04:17<17:36,  4.38s/it]
 20%|██        | 60/300 [04:22<17:32,  4.38s/it]
                                                
 20%|██        | 60/300 [04:22<17:32,  4.38s/it]
 20%|██        | 61/300 [04:26<17:29,  4.39s/it]
 21%|██        | 62/300 [04:31<18:03,  4.55s/it]
 21%|██        | 63/300 [04:35<17:48,  4.51s/it]
 21%|██▏       | 64/300 [04:40<17:32,  4.46s/it]
 22%|██▏       | 65/300 [04:44<17:18,  4.42s/it]
 22%|██▏       | 66/300 [04:48<17:06,  4.39s/it]
 22%|██▏       | 67/300 [04:53<17:02,  4.39s/it]
 23%|██▎       | 68/300 [04:57<16:45,  4.33s/it]
 23%|██▎       | 69/300 [05:01<16:44,  4.35s/it]
 23%|██▎       | 70/300 [05:06<16:30,  4.31s/it]
 24%|██▎       | 71/300 [05:10<16:30,  4.33s/it]
 24%|██▍       | 72/300 [05:14<16:29,  4.34s/it]
 24%|██▍       | 73/300 [05:19<16:28,  4.36s/it]
 25%|██▍       | 74/300 [05:23<16:24,  4.36s/it]
 25%|██▌       | 75/300 [05:28<16:57,  4.52s/it]
 25%|██▌       | 76/300 [05:32<16:43,  4.48s/it]
 26%|██▌       | 77/300 [05:37<16:30,  4.44s/it]
 26%|██▌       | 78/300 [05:41<16:17,  4.40s/it]
 26%|██▋       | 79/300 [05:45<16:02,  4.36s/it]
 27%|██▋       | 80/300 [05:50<16:00,  4.37s/it]
                                                
 27%|██▋       | 80/300 [05:50<16:00,  4.37s/it]
 27%|██▋       | 81/300 [05:54<15:54,  4.36s/it]
 27%|██▋       | 82/300 [05:58<15:51,  4.36s/it]
 28%|██▊       | 83/300 [06:03<15:35,  4.31s/it]
 28%|██▊       | 84/300 [06:07<15:32,  4.32s/it]
 28%|██▊       | 85/300 [06:11<15:31,  4.33s/it]
 29%|██▊       | 86/300 [06:16<15:26,  4.33s/it]
 29%|██▉       | 87/300 [06:20<15:04,  4.25s/it]
 29%|██▉       | 88/300 [06:24<15:03,  4.26s/it]
 30%|██▉       | 89/300 [06:28<15:04,  4.29s/it]
 30%|███       | 90/300 [06:33<15:08,  4.33s/it]
 30%|███       | 91/300 [06:37<15:06,  4.34s/it]
 31%|███       | 92/300 [06:42<15:43,  4.54s/it]
 31%|███       | 93/300 [06:46<15:19,  4.44s/it]
 31%|███▏      | 94/300 [06:51<15:10,  4.42s/it]
 32%|███▏      | 95/300 [06:55<15:02,  4.40s/it]
 32%|███▏      | 96/300 [06:59<14:45,  4.34s/it]
 32%|███▏      | 97/300 [07:04<14:44,  4.36s/it]
 33%|███▎      | 98/300 [07:08<14:40,  4.36s/it]
 33%|███▎      | 99/300 [07:12<14:30,  4.33s/it]
 33%|███▎      | 100/300 [07:17<14:23,  4.32s/it]
                                                 
 33%|███▎      | 100/300 [07:17<14:23,  4.32s/it]
 34%|███▎      | 101/300 [07:21<14:16,  4.30s/it]
 34%|███▍      | 102/300 [07:25<14:14,  4.32s/it]
 34%|███▍      | 103/300 [07:30<14:12,  4.33s/it]
 35%|███▍      | 104/300 [07:34<14:00,  4.29s/it]
 35%|███▌      | 105/300 [07:38<14:01,  4.32s/it]
 35%|███▌      | 106/300 [07:42<13:59,  4.33s/it]
 36%|███▌      | 107/300 [07:47<14:30,  4.51s/it]
 36%|███▌      | 108/300 [07:52<14:16,  4.46s/it]
 36%|███▋      | 109/300 [07:56<14:07,  4.44s/it]
 37%|███▋      | 110/300 [08:00<14:00,  4.42s/it]
 37%|███▋      | 111/300 [08:05<13:53,  4.41s/it]
 37%|███▋      | 112/300 [08:09<13:47,  4.40s/it]
 38%|███▊      | 113/300 [08:14<13:39,  4.38s/it]
 38%|███▊      | 114/300 [08:18<13:32,  4.37s/it]
 38%|███▊      | 115/300 [08:22<13:27,  4.37s/it]
 39%|███▊      | 116/300 [08:27<13:24,  4.37s/it]
 39%|███▉      | 117/300 [08:31<13:20,  4.38s/it]
 39%|███▉      | 118/300 [08:35<13:12,  4.36s/it]
 40%|███▉      | 119/300 [08:40<13:09,  4.36s/it]
 40%|████      | 120/300 [08:44<13:07,  4.38s/it]
                                                 
 40%|████      | 120/300 [08:44<13:07,  4.38s/it]
 40%|████      | 121/300 [08:49<13:03,  4.38s/it]
 41%|████      | 122/300 [08:54<13:30,  4.55s/it]
 41%|████      | 123/300 [08:58<13:17,  4.51s/it]
 41%|████▏     | 124/300 [09:02<13:06,  4.47s/it]
 42%|████▏     | 125/300 [09:07<12:57,  4.44s/it]
 42%|████▏     | 126/300 [09:11<12:50,  4.43s/it]
 42%|████▏     | 127/300 [09:15<12:42,  4.41s/it]
 43%|████▎     | 128/300 [09:20<12:34,  4.39s/it]
 43%|████▎     | 129/300 [09:24<12:27,  4.37s/it]
 43%|████▎     | 130/300 [09:28<12:19,  4.35s/it]
 44%|████▎     | 131/300 [09:33<12:17,  4.37s/it]
 44%|████▍     | 132/300 [09:37<12:14,  4.37s/it]
 44%|████▍     | 133/300 [09:41<12:06,  4.35s/it]
 45%|████▍     | 134/300 [09:46<12:02,  4.35s/it]
 45%|████▌     | 135/300 [09:50<11:58,  4.36s/it]
 45%|████▌     | 136/300 [09:55<12:22,  4.53s/it]
 46%|████▌     | 137/300 [10:00<12:10,  4.48s/it]
 46%|████▌     | 138/300 [10:04<12:01,  4.46s/it]
 46%|████▋     | 139/300 [10:08<11:55,  4.44s/it]
 47%|████▋     | 140/300 [10:13<11:45,  4.41s/it]
                                                 
 47%|████▋     | 140/300 [10:13<11:45,  4.41s/it]
 47%|████▋     | 141/300 [10:17<11:37,  4.39s/it]
 47%|████▋     | 142/300 [10:21<11:28,  4.36s/it]
 48%|████▊     | 143/300 [10:26<11:22,  4.35s/it]
 48%|████▊     | 144/300 [10:30<11:13,  4.31s/it]
 48%|████▊     | 145/300 [10:34<11:09,  4.32s/it]
 49%|████▊     | 146/300 [10:39<11:07,  4.34s/it]
 49%|████▉     | 147/300 [10:43<11:07,  4.37s/it]
 49%|████▉     | 148/300 [10:47<11:05,  4.38s/it]
 50%|████▉     | 149/300 [10:52<10:57,  4.35s/it]
 50%|█████     | 150/300 [10:56<10:53,  4.36s/it]
 50%|█████     | 151/300 [11:00<10:49,  4.36s/it]
 51%|█████     | 152/300 [11:05<11:10,  4.53s/it]
 51%|█████     | 153/300 [11:10<10:54,  4.46s/it]
 51%|█████▏    | 154/300 [11:14<10:41,  4.39s/it]
 52%|█████▏    | 155/300 [11:18<10:35,  4.39s/it]
 52%|█████▏    | 156/300 [11:23<10:30,  4.38s/it]
 52%|█████▏    | 157/300 [11:27<10:24,  4.37s/it]
 53%|█████▎    | 158/300 [11:31<10:18,  4.35s/it]
 53%|█████▎    | 159/300 [11:36<10:14,  4.36s/it]
 53%|█████▎    | 160/300 [11:40<10:09,  4.35s/it]
                                                 
 53%|█████▎    | 160/300 [11:40<10:09,  4.35s/it]
 54%|█████▎    | 161/300 [11:44<10:02,  4.33s/it]
 54%|█████▍    | 162/300 [11:49<09:59,  4.35s/it]
 54%|█████▍    | 163/300 [11:53<09:55,  4.35s/it]
 55%|█████▍    | 164/300 [11:57<09:53,  4.36s/it]
 55%|█████▌    | 165/300 [12:02<09:50,  4.37s/it]
 55%|█████▌    | 166/300 [12:06<09:46,  4.38s/it]
 56%|█████▌    | 167/300 [12:11<09:43,  4.39s/it]
 56%|█████▌    | 168/300 [12:15<09:58,  4.53s/it]
 56%|█████▋    | 169/300 [12:20<09:45,  4.47s/it]
 57%|█████▋    | 170/300 [12:24<09:36,  4.44s/it]
 57%|█████▋    | 171/300 [12:28<09:28,  4.41s/it]
 57%|█████▋    | 172/300 [12:33<09:23,  4.40s/it]
 58%|█████▊    | 173/300 [12:37<09:18,  4.40s/it]
 58%|█████▊    | 174/300 [12:42<09:10,  4.37s/it]
 58%|█████▊    | 175/300 [12:46<09:05,  4.36s/it]
 59%|█████▊    | 176/300 [12:50<08:55,  4.32s/it]
 59%|█████▉    | 177/300 [12:54<08:40,  4.23s/it]
 59%|█████▉    | 178/300 [12:58<08:40,  4.27s/it]
 60%|█████▉    | 179/300 [13:03<08:37,  4.28s/it]
 60%|██████    | 180/300 [13:07<08:37,  4.31s/it]
                                                 
 60%|██████    | 180/300 [13:07<08:37,  4.31s/it]
 60%|██████    | 181/300 [13:12<08:33,  4.32s/it]
 61%|██████    | 182/300 [13:16<08:32,  4.34s/it]
 61%|██████    | 183/300 [13:21<08:47,  4.50s/it]
 61%|██████▏   | 184/300 [13:25<08:34,  4.43s/it]
 62%|██████▏   | 185/300 [13:29<08:22,  4.37s/it]
 62%|██████▏   | 186/300 [13:34<08:20,  4.39s/it]
 62%|██████▏   | 187/300 [13:38<08:14,  4.38s/it]
 63%|██████▎   | 188/300 [13:42<08:09,  4.37s/it]
 63%|██████▎   | 189/300 [13:47<07:57,  4.30s/it]
 63%|██████▎   | 190/300 [13:51<07:54,  4.31s/it]
 64%|██████▎   | 191/300 [13:55<07:54,  4.35s/it]
 64%|██████▍   | 192/300 [14:00<07:48,  4.34s/it]
 64%|██████▍   | 193/300 [14:04<07:43,  4.33s/it]
 65%|██████▍   | 194/300 [14:08<07:40,  4.34s/it]
 65%|██████▌   | 195/300 [14:13<07:38,  4.36s/it]
 65%|██████▌   | 196/300 [14:17<07:34,  4.37s/it]
 66%|██████▌   | 197/300 [14:22<07:30,  4.38s/it]
 66%|██████▌   | 198/300 [14:26<07:42,  4.53s/it]
 66%|██████▋   | 199/300 [14:31<07:34,  4.50s/it]
 67%|██████▋   | 200/300 [14:35<07:24,  4.44s/it]
                                                 
 67%|██████▋   | 200/300 [14:35<07:24,  4.44s/it]
 67%|██████▋   | 201/300 [14:40<07:17,  4.42s/it]
 67%|██████▋   | 202/300 [14:44<07:10,  4.39s/it]
 68%|██████▊   | 203/300 [14:48<07:07,  4.41s/it]
 68%|██████▊   | 204/300 [14:53<07:02,  4.40s/it]
 68%|██████▊   | 205/300 [14:57<06:58,  4.41s/it]
 69%|██████▊   | 206/300 [15:01<06:53,  4.40s/it]
 69%|██████▉   | 207/300 [15:06<06:48,  4.39s/it]
 69%|██████▉   | 208/300 [15:10<06:42,  4.37s/it]
 70%|██████▉   | 209/300 [15:15<06:38,  4.38s/it]
 70%|███████   | 210/300 [15:19<06:35,  4.39s/it]
 70%|███████   | 211/300 [15:23<06:29,  4.37s/it]
 71%|███████   | 212/300 [15:27<06:19,  4.31s/it]
 71%|███████   | 213/300 [15:32<06:32,  4.51s/it]
 71%|███████▏  | 214/300 [15:37<06:21,  4.44s/it]
 72%|███████▏  | 215/300 [15:41<06:15,  4.42s/it]
 72%|███████▏  | 216/300 [15:45<06:05,  4.36s/it]
 72%|███████▏  | 217/300 [15:50<05:58,  4.32s/it]
 73%|███████▎  | 218/300 [15:54<05:56,  4.35s/it]
 73%|███████▎  | 219/300 [15:58<05:53,  4.36s/it]
 73%|███████▎  | 220/300 [16:03<05:49,  4.37s/it]
                                                 
 73%|███████▎  | 220/300 [16:03<05:49,  4.37s/it]
 74%|███████▎  | 221/300 [16:07<05:45,  4.37s/it]
 74%|███████▍  | 222/300 [16:11<05:40,  4.37s/it]
 74%|███████▍  | 223/300 [16:16<05:36,  4.37s/it]
 75%|███████▍  | 224/300 [16:20<05:29,  4.34s/it]
 75%|███████▌  | 225/300 [16:24<05:23,  4.31s/it]
 75%|███████▌  | 226/300 [16:29<05:21,  4.34s/it]
 76%|███████▌  | 227/300 [16:33<05:12,  4.29s/it]
 76%|███████▌  | 228/300 [16:37<05:10,  4.32s/it]
 76%|███████▋  | 229/300 [16:42<05:05,  4.31s/it]
 77%|███████▋  | 230/300 [16:46<05:06,  4.39s/it]
 77%|███████▋  | 231/300 [16:51<05:01,  4.37s/it]
 77%|███████▋  | 232/300 [16:55<04:58,  4.38s/it]
 78%|███████▊  | 233/300 [16:59<04:51,  4.35s/it]
 78%|███████▊  | 234/300 [17:04<04:47,  4.36s/it]
 78%|███████▊  | 235/300 [17:08<04:44,  4.37s/it]
 79%|███████▊  | 236/300 [17:12<04:38,  4.36s/it]
 79%|███████▉  | 237/300 [17:17<04:34,  4.36s/it]
 79%|███████▉  | 238/300 [17:21<04:29,  4.35s/it]
 80%|███████▉  | 239/300 [17:25<04:24,  4.33s/it]
 80%|████████  | 240/300 [17:30<04:20,  4.34s/it]
                                                 
 80%|████████  | 240/300 [17:30<04:20,  4.34s/it]
 80%|████████  | 241/300 [17:34<04:16,  4.36s/it]
 81%|████████  | 242/300 [17:38<04:12,  4.35s/it]
 81%|████████  | 243/300 [17:43<04:06,  4.33s/it]
 81%|████████▏ | 244/300 [17:47<03:59,  4.27s/it]
 82%|████████▏ | 245/300 [17:51<04:01,  4.40s/it]
 82%|████████▏ | 246/300 [17:56<03:57,  4.40s/it]
 82%|████████▏ | 247/300 [18:00<03:52,  4.38s/it]
 83%|████████▎ | 248/300 [18:05<03:46,  4.36s/it]
 83%|████████▎ | 249/300 [18:09<03:42,  4.36s/it]
 83%|████████▎ | 250/300 [18:13<03:38,  4.37s/it]
 84%|████████▎ | 251/300 [18:17<03:31,  4.32s/it]
 84%|████████▍ | 252/300 [18:22<03:28,  4.34s/it]
 84%|████████▍ | 253/300 [18:26<03:25,  4.36s/it]
 85%|████████▍ | 254/300 [18:31<03:20,  4.37s/it]
 85%|████████▌ | 255/300 [18:35<03:15,  4.35s/it]
 85%|████████▌ | 256/300 [18:39<03:10,  4.34s/it]
 86%|████████▌ | 257/300 [18:44<03:07,  4.36s/it]
 86%|████████▌ | 258/300 [18:48<03:02,  4.35s/it]
 86%|████████▋ | 259/300 [18:53<03:05,  4.52s/it]
 87%|████████▋ | 260/300 [18:57<02:59,  4.48s/it]
                                                 
 87%|████████▋ | 260/300 [18:57<02:59,  4.48s/it]
 87%|████████▋ | 261/300 [19:02<02:52,  4.42s/it]
 87%|████████▋ | 262/300 [19:06<02:47,  4.41s/it]
 88%|████████▊ | 263/300 [19:10<02:41,  4.36s/it]
 88%|████████▊ | 264/300 [19:15<02:36,  4.33s/it]
 88%|████████▊ | 265/300 [19:19<02:31,  4.33s/it]
 89%|████████▊ | 266/300 [19:23<02:27,  4.35s/it]
 89%|████████▉ | 267/300 [19:28<02:23,  4.35s/it]
 89%|████████▉ | 268/300 [19:32<02:19,  4.37s/it]
 90%|████████▉ | 269/300 [19:36<02:14,  4.35s/it]
 90%|█████████ | 270/300 [19:41<02:10,  4.35s/it]
 90%|█████████ | 271/300 [19:45<02:06,  4.36s/it]
 91%|█████████ | 272/300 [19:49<02:01,  4.35s/it]
 91%|█████████ | 273/300 [19:54<02:02,  4.52s/it]
 91%|█████████▏| 274/300 [19:59<01:56,  4.47s/it]
 92%|█████████▏| 275/300 [20:03<01:51,  4.45s/it]
 92%|█████████▏| 276/300 [20:07<01:46,  4.44s/it]
 92%|█████████▏| 277/300 [20:12<01:40,  4.37s/it]
 93%|█████████▎| 278/300 [20:16<01:35,  4.34s/it]
 93%|█████████▎| 279/300 [20:20<01:31,  4.34s/it]
 93%|█████████▎| 280/300 [20:24<01:25,  4.29s/it]
                                                 
 93%|█████████▎| 280/300 [20:24<01:25,  4.29s/it]
 94%|█████████▎| 281/300 [20:29<01:21,  4.30s/it]
 94%|█████████▍| 282/300 [20:33<01:17,  4.32s/it]
 94%|█████████▍| 283/300 [20:38<01:13,  4.35s/it]
 95%|█████████▍| 284/300 [20:42<01:09,  4.34s/it]
 95%|█████████▌| 285/300 [20:46<01:05,  4.35s/it]
 95%|█████████▌| 286/300 [20:51<01:00,  4.32s/it]
 96%|█████████▌| 287/300 [20:55<00:55,  4.29s/it]
 96%|█████████▌| 288/300 [20:59<00:51,  4.32s/it]
 96%|█████████▋| 289/300 [21:03<00:47,  4.31s/it]
 97%|█████████▋| 290/300 [21:08<00:44,  4.44s/it]
 97%|█████████▋| 291/300 [21:12<00:39,  4.39s/it]
 97%|█████████▋| 292/300 [21:17<00:35,  4.38s/it]
 98%|█████████▊| 293/300 [21:21<00:30,  4.38s/it]
 98%|█████████▊| 294/300 [21:25<00:26,  4.35s/it]
 98%|█████████▊| 295/300 [21:30<00:21,  4.30s/it]
 99%|█████████▊| 296/300 [21:34<00:17,  4.31s/it]
 99%|█████████▉| 297/300 [21:38<00:13,  4.34s/it]
 99%|█████████▉| 298/300 [21:43<00:08,  4.33s/it]
                                                 
+  warnings.warn(
+
                                                 
+{'loss': 0.3144, 'grad_norm': 0.21818794310092926, 'learning_rate': 0.00019941781378123244, 'epoch': 0.01}
+{'loss': 0.2763, 'grad_norm': 0.174688920378685, 'learning_rate': 0.00019480091799562704, 'epoch': 0.02}
+{'loss': 0.2686, 'grad_norm': 0.20462645590305328, 'learning_rate': 0.00018578153193067745, 'epoch': 0.03}
+{'loss': 0.2772, 'grad_norm': 0.18597012758255005, 'learning_rate': 0.00017277850975273696, 'epoch': 0.04}
+{'loss': 0.2715, 'grad_norm': 0.17670691013336182, 'learning_rate': 0.00015639570298509064, 'epoch': 0.05}
+{'loss': 0.2809, 'grad_norm': 0.18298876285552979, 'learning_rate': 0.00013739391805311793, 'epoch': 0.07}
+{'loss': 0.2823, 'grad_norm': 0.1972775161266327, 'learning_rate': 0.00011665558495034546, 'epoch': 0.08}
+{'loss': 0.2673, 'grad_norm': 0.1763097494840622, 'learning_rate': 9.514377778855521e-05, 'epoch': 0.09}
+{'loss': 0.2734, 'grad_norm': 0.15922492742538452, 'learning_rate': 7.385749028978346e-05, 'epoch': 0.1}
+{'loss': 0.27, 'grad_norm': 0.16574564576148987, 'learning_rate': 5.378524319600231e-05, 'epoch': 0.11}
+{'loss': 0.2631, 'grad_norm': 0.1575671285390854, 'learning_rate': 3.585917803688603e-05, 'epoch': 0.12}
+{'loss': 0.2633, 'grad_norm': 0.17549917101860046, 'learning_rate': 2.091176910992545e-05, 'epoch': 0.13}
+{'loss': 0.2594, 'grad_norm': 0.17588092386722565, 'learning_rate': 9.637163939149485e-06, 'epoch': 0.14}
+{'loss': 0.2645, 'grad_norm': 0.17386789619922638, 'learning_rate': 2.5589475353073988e-06, 'epoch': 0.15}
+{'loss': 0.2748, 'grad_norm': 0.17659908533096313, 'learning_rate': 5.827463387653165e-09, 'epoch': 0.16}
+{'train_runtime': 1314.1404, 'train_samples_per_second': 3.653, 'train_steps_per_second': 0.228, 'train_loss': 0.27380541801452635, 'epoch': 0.16}
+train time (s): 1315.014744758606
+/usr/local/lib/python3.10/dist-packages/peft/utils/save_and_load.py:295: UserWarning: Could not find a config file in Qwen/Qwen3-4B-Base - will assume that the vocabulary was not modified.
+  warnings.warn(
+saved to out/lora_probe

instances/end_of_run/d05/state/out/main1.log

@@ -0,0 +1,48 @@
+⚙️  Running in WANDB offline mode
+training rows: 64247
+pretokenized in 71.1s -> 29238 sequences
+`torch_dtype` is deprecated! Use `dtype` instead!
+
+trainable params: 132,120,576 || all params: 4,154,588,672 || trainable%: 3.1801
+
  0%|          | 0/600 [00:00<?, ?it/s]
  0%|          | 1/600 [00:07<1:12:36,  7.27s/it]
  0%|          | 2/600 [00:11<54:43,  5.49s/it]  
  0%|          | 3/600 [00:15<49:14,  4.95s/it]
  1%|          | 4/600 [00:20<46:25,  4.67s/it]
  1%|          | 5/600 [00:24<44:48,  4.52s/it]
  1%|          | 6/600 [00:28<44:15,  4.47s/it]
  1%|          | 7/600 [00:32<43:19,  4.38s/it]
  1%|▏         | 8/600 [00:37<43:08,  4.37s/it]
  2%|▏         | 9/600 [00:41<42:59,  4.37s/it]
  2%|▏         | 10/600 [00:45<42:34,  4.33s/it]
  2%|▏         | 11/600 [00:50<42:27,  4.33s/it]
  2%|▏         | 12/600 [00:54<42:19,  4.32s/it]
  2%|▏         | 13/600 [00:58<41:39,  4.26s/it]
  2%|▏         | 14/600 [01:02<41:41,  4.27s/it]
  2%|▎         | 15/600 [01:07<42:00,  4.31s/it]
  3%|▎         | 16/600 [01:12<43:50,  4.50s/it]
  3%|▎         | 17/600 [01:16<43:04,  4.43s/it]
  3%|▎         | 18/600 [01:20<42:47,  4.41s/it]
  3%|▎         | 19/600 [01:25<42:05,  4.35s/it]
  3%|▎         | 20/600 [01:29<42:04,  4.35s/it]
                                                
  3%|▎         | 20/600 [01:29<42:04,  4.35s/it]
  4%|▎         | 21/600 [01:33<42:10,  4.37s/it]
  4%|▎         | 22/600 [01:38<42:06,  4.37s/it]
  4%|▍         | 23/600 [01:42<42:05,  4.38s/it]
  4%|▍         | 24/600 [01:46<41:30,  4.32s/it]
  4%|▍         | 25/600 [01:51<41:29,  4.33s/it]
  4%|▍         | 26/600 [01:55<41:15,  4.31s/it]
  4%|▍         | 27/600 [01:59<41:25,  4.34s/it]
  5%|▍         | 28/600 [02:04<41:28,  4.35s/it]
  5%|▍         | 29/600 [02:08<41:30,  4.36s/it]
  5%|▌         | 30/600 [02:12<41:21,  4.35s/it]
  5%|▌         | 31/600 [02:17<41:20,  4.36s/it]
  5%|▌         | 32/600 [02:21<41:14,  4.36s/it]
  6%|▌         | 33/600 [02:26<42:35,  4.51s/it]
  6%|▌         | 34/600 [02:30<42:06,  4.46s/it]
  6%|▌         | 35/600 [02:34<40:47,  4.33s/it]
  6%|▌         | 36/600 [02:39<40:40,  4.33s/it]
  6%|▌         | 37/600 [02:43<40:44,  4.34s/it]
  6%|▋         | 38/600 [02:47<40:48,  4.36s/it]
  6%|▋         | 39/600 [02:52<40:18,  4.31s/it]
  7%|▋         | 40/600 [02:56<40:31,  4.34s/it]
                                                
  7%|▋         | 40/600 [02:56<40:31,  4.34s/it]
  7%|▋         | 41/600 [03:00<40:19,  4.33s/it]
  7%|▋         | 42/600 [03:05<40:24,  4.35s/it]
  7%|▋         | 43/600 [03:09<40:19,  4.34s/it]
  7%|▋         | 44/600 [03:14<40:24,  4.36s/it]
  8%|▊         | 45/600 [03:18<40:20,  4.36s/it]
  8%|▊         | 46/600 [03:22<40:27,  4.38s/it]
  8%|▊         | 47/600 [03:27<40:21,  4.38s/it]
  8%|▊         | 48/600 [03:31<41:27,  4.51s/it]
  8%|▊         | 49/600 [03:36<41:04,  4.47s/it]
  8%|▊         | 50/600 [03:40<40:51,  4.46s/it]
  8%|▊         | 51/600 [03:45<40:25,  4.42s/it]
  9%|▊         | 52/600 [03:49<39:50,  4.36s/it]
  9%|▉         | 53/600 [03:53<39:33,  4.34s/it]
  9%|▉         | 54/600 [03:57<39:12,  4.31s/it]
  9%|▉         | 55/600 [04:02<39:22,  4.33s/it]
  9%|▉         | 56/600 [04:06<39:28,  4.35s/it]
 10%|▉         | 57/600 [04:11<39:28,  4.36s/it]
 10%|▉         | 58/600 [04:15<39:22,  4.36s/it]
 10%|▉         | 59/600 [04:19<39:28,  4.38s/it]
 10%|█         | 60/600 [04:24<39:06,  4.34s/it]
                                                
 10%|█         | 60/600 [04:24<39:06,  4.34s/it]
 10%|█         | 61/600 [04:28<39:13,  4.37s/it]
 10%|█         | 62/600 [04:33<40:39,  4.53s/it]
 10%|█         | 63/600 [04:37<40:10,  4.49s/it]
 11%|█         | 64/600 [04:42<39:24,  4.41s/it]
 11%|█         | 65/600 [04:46<39:24,  4.42s/it]
 11%|█         | 66/600 [04:50<38:51,  4.37s/it]
 11%|█         | 67/600 [04:55<38:53,  4.38s/it]
 11%|█▏        | 68/600 [04:59<38:55,  4.39s/it]
 12%|█▏        | 69/600 [05:03<38:44,  4.38s/it]
 12%|█▏        | 70/600 [05:08<38:36,  4.37s/it]
 12%|█▏        | 71/600 [05:12<38:32,  4.37s/it]
 12%|█▏        | 72/600 [05:16<38:21,  4.36s/it]
 12%|█▏        | 73/600 [05:21<38:26,  4.38s/it]
 12%|█▏        | 74/600 [05:25<38:07,  4.35s/it]
 12%|█▎        | 75/600 [05:30<39:29,  4.51s/it]
 13%|█▎        | 76/600 [05:34<38:39,  4.43s/it]
 13%|█▎        | 77/600 [05:39<38:28,  4.41s/it]
 13%|█▎        | 78/600 [05:43<38:21,  4.41s/it]
 13%|█▎        | 79/600 [05:47<37:52,  4.36s/it]
 13%|█▎        | 80/600 [05:52<37:55,  4.38s/it]
                                                
 13%|█▎        | 80/600 [05:52<37:55,  4.38s/it]
 14%|█▎        | 81/600 [05:56<36:47,  4.25s/it]
 14%|█▎        | 82/600 [06:00<36:58,  4.28s/it]
 14%|█▍        | 83/600 [06:04<36:45,  4.27s/it]
 14%|█▍        | 84/600 [06:09<36:59,  4.30s/it]
 14%|█▍        | 85/600 [06:13<37:12,  4.33s/it]
 14%|█▍        | 86/600 [06:17<37:04,  4.33s/it]
 14%|█▍        | 87/600 [06:22<37:04,  4.34s/it]
 15%|█▍        | 88/600 [06:26<37:12,  4.36s/it]
 15%|█▍        | 89/600 [06:31<37:15,  4.37s/it]
 15%|█▌        | 90/600 [06:35<37:12,  4.38s/it]
 15%|█▌        | 91/600 [06:39<37:12,  4.39s/it]
 15%|█▌        | 92/600 [06:44<38:35,  4.56s/it]
 16%|█▌        | 93/600 [06:48<37:34,  4.45s/it]
 16%|█▌        | 94/600 [06:53<37:16,  4.42s/it]
 16%|█▌        | 95/600 [06:57<37:05,  4.41s/it]
 16%|█▌        | 96/600 [07:01<36:34,  4.35s/it]
 16%|█▌        | 97/600 [07:06<36:40,  4.37s/it]
 16%|█▋        | 98/600 [07:10<36:32,  4.37s/it]
 16%|█▋        | 99/600 [07:15<36:34,  4.38s/it]
 17%|█▋        | 100/600 [07:19<36:35,  4.39s/it]
                                                 
 17%|█▋        | 100/600 [07:19<36:35,  4.39s/it]
 17%|█▋        | 101/600 [07:23<36:31,  4.39s/it]
 17%|█▋        | 102/600 [07:28<36:22,  4.38s/it]
 17%|█▋        | 103/600 [07:32<36:28,  4.40s/it]
 17%|█▋        | 104/600 [07:37<36:26,  4.41s/it]
 18%|█▊        | 105/600 [07:41<35:52,  4.35s/it]
 18%|█▊        | 106/600 [07:45<35:50,  4.35s/it]
 18%|█▊        | 107/600 [07:50<37:16,  4.54s/it]
 18%|█▊        | 108/600 [07:55<36:46,  4.49s/it]
 18%|█▊        | 109/600 [07:59<36:21,  4.44s/it]
 18%|█▊        | 110/600 [08:03<36:11,  4.43s/it]
 18%|█▊        | 111/600 [08:08<35:57,  4.41s/it]
 19%|█▊        | 112/600 [08:12<35:45,  4.40s/it]
 19%|█▉        | 113/600 [08:16<35:44,  4.40s/it]
 19%|█▉        | 114/600 [08:21<35:39,  4.40s/it]
 19%|█▉        | 115/600 [08:25<35:39,  4.41s/it]
 19%|█▉        | 116/600 [08:30<35:36,  4.41s/it]
 20%|█▉        | 117/600 [08:34<35:35,  4.42s/it]
 20%|█▉        | 118/600 [08:39<35:22,  4.40s/it]
 20%|█▉        | 119/600 [08:43<35:15,  4.40s/it]
 20%|██        | 120/600 [08:47<35:04,  4.38s/it]
                                                 
 20%|██        | 120/600 [08:47<35:04,  4.38s/it]
 20%|██        | 121/600 [08:52<35:07,  4.40s/it]
 20%|██        | 122/600 [08:57<36:27,  4.58s/it]
 20%|██        | 123/600 [09:01<35:58,  4.53s/it]
 21%|██        | 124/600 [09:06<35:42,  4.50s/it]
 21%|██        | 125/600 [09:10<35:14,  4.45s/it]
 21%|██        | 126/600 [09:14<34:45,  4.40s/it]
 21%|██        | 127/600 [09:19<34:41,  4.40s/it]
 21%|██▏       | 128/600 [09:23<34:44,  4.42s/it]
 22%|██▏       | 129/600 [09:27<34:45,  4.43s/it]
 22%|██▏       | 130/600 [09:32<34:35,  4.42s/it]
 22%|██▏       | 131/600 [09:36<34:18,  4.39s/it]
 22%|██▏       | 132/600 [09:40<33:56,  4.35s/it]
 22%|██▏       | 133/600 [09:45<33:49,  4.35s/it]
 22%|██▏       | 134/600 [09:49<33:47,  4.35s/it]
 22%|██▎       | 135/600 [09:53<33:42,  4.35s/it]
 23%|██▎       | 136/600 [09:58<35:06,  4.54s/it]
 23%|██▎       | 137/600 [10:03<34:19,  4.45s/it]
 23%|██▎       | 138/600 [10:07<33:53,  4.40s/it]
 23%|██▎       | 139/600 [10:11<33:50,  4.41s/it]
 23%|██▎       | 140/600 [10:16<33:50,  4.41s/it]
                                                 
 23%|██▎       | 140/600 [10:16<33:50,  4.41s/it]
 24%|██▎       | 141/600 [10:20<33:32,  4.38s/it]
 24%|██▎       | 142/600 [10:25<33:30,  4.39s/it]
 24%|██▍       | 143/600 [10:29<33:28,  4.39s/it]
 24%|██▍       | 144/600 [10:33<33:27,  4.40s/it]
 24%|██▍       | 145/600 [10:38<33:22,  4.40s/it]
 24%|██▍       | 146/600 [10:42<33:16,  4.40s/it]
 24%|██▍       | 147/600 [10:47<33:12,  4.40s/it]
 25%|██▍       | 148/600 [10:51<33:00,  4.38s/it]
 25%|██▍       | 149/600 [10:55<32:58,  4.39s/it]
 25%|██▌       | 150/600 [11:00<32:48,  4.38s/it]
 25%|██▌       | 151/600 [11:04<32:41,  4.37s/it]
 25%|██▌       | 152/600 [11:09<33:39,  4.51s/it]
 26%|██▌       | 153/600 [11:13<33:19,  4.47s/it]
 26%|██▌       | 154/600 [11:18<32:57,  4.43s/it]
 26%|██▌       | 155/600 [11:22<32:53,  4.43s/it]
 26%|██▌       | 156/600 [11:26<32:33,  4.40s/it]
 26%|██▌       | 157/600 [11:31<32:35,  4.41s/it]
 26%|██▋       | 158/600 [11:35<32:26,  4.40s/it]
 26%|██▋       | 159/600 [11:39<31:56,  4.34s/it]
 27%|██▋       | 160/600 [11:44<31:57,  4.36s/it]
                                                 
 27%|██▋       | 160/600 [11:44<31:57,  4.36s/it]
 27%|██▋       | 161/600 [11:48<31:52,  4.36s/it]
 27%|██▋       | 162/600 [11:53<31:51,  4.37s/it]
 27%|██▋       | 163/600 [11:57<31:26,  4.32s/it]
 27%|██▋       | 164/600 [12:01<31:27,  4.33s/it]
 28%|██▊       | 165/600 [12:05<31:30,  4.35s/it]
 28%|██▊       | 166/600 [12:10<31:25,  4.34s/it]
 28%|██▊       | 167/600 [12:14<31:22,  4.35s/it]
 28%|██▊       | 168/600 [12:19<32:23,  4.50s/it]
 28%|██▊       | 169/600 [12:23<31:57,  4.45s/it]
 28%|██▊       | 170/600 [12:28<31:40,  4.42s/it]
 28%|██▊       | 171/600 [12:32<31:07,  4.35s/it]
 29%|██▊       | 172/600 [12:36<31:09,  4.37s/it]
 29%|██▉       | 173/600 [12:40<30:45,  4.32s/it]
 29%|██▉       | 174/600 [12:45<30:14,  4.26s/it]
 29%|██▉       | 175/600 [12:49<30:31,  4.31s/it]
 29%|██▉       | 176/600 [12:53<30:43,  4.35s/it]
 30%|██▉       | 177/600 [12:58<30:33,  4.34s/it]
 30%|██▉       | 178/600 [13:02<30:37,  4.35s/it]
 30%|██▉       | 179/600 [13:07<30:35,  4.36s/it]
 30%|███       | 180/600 [13:11<30:41,  4.38s/it]
                                                 
 30%|███       | 180/600 [13:11<30:41,  4.38s/it]
 30%|███       | 181/600 [13:15<30:38,  4.39s/it]
 30%|███       | 182/600 [13:20<30:39,  4.40s/it]
 30%|███       | 183/600 [13:25<31:43,  4.56s/it]
 31%|███       | 184/600 [13:29<30:51,  4.45s/it]
 31%|███       | 185/600 [13:33<30:40,  4.44s/it]
 31%|███       | 186/600 [13:38<30:30,  4.42s/it]
 31%|███       | 187/600 [13:42<30:25,  4.42s/it]
 31%|███▏      | 188/600 [13:46<29:57,  4.36s/it]
 32%|███▏      | 189/600 [13:51<29:58,  4.37s/it]
 32%|███▏      | 190/600 [13:55<29:40,  4.34s/it]
 32%|███▏      | 191/600 [13:59<29:45,  4.36s/it]
 32%|███▏      | 192/600 [14:04<29:51,  4.39s/it]
 32%|███▏      | 193/600 [14:08<29:37,  4.37s/it]
 32%|███▏      | 194/600 [14:13<29:36,  4.37s/it]
 32%|███▎      | 195/600 [14:17<29:06,  4.31s/it]
 33%|███▎      | 196/600 [14:21<28:46,  4.27s/it]
 33%|███▎      | 197/600 [14:25<28:54,  4.31s/it]
 33%|███▎      | 198/600 [14:30<30:15,  4.52s/it]
 33%|███▎      | 199/600 [14:35<30:00,  4.49s/it]
 33%|███▎      | 200/600 [14:39<29:48,  4.47s/it]
                                                 
 33%|███▎      | 200/600 [14:39<29:48,  4.47s/it]/usr/local/lib/python3.10/dist-packages/peft/utils/save_and_load.py:295: UserWarning: Could not find a config file in Qwen/Qwen3-4B-Base - will assume that the vocabulary was not modified.
+  warnings.warn(
+
 34%|███▎      | 201/600 [14:46<34:45,  5.23s/it]
 34%|███▎      | 202/600 [14:51<33:05,  4.99s/it]
 34%|███▍      | 203/600 [14:55<31:29,  4.76s/it]
 34%|███▍      | 204/600 [14:59<30:36,  4.64s/it]
 34%|███▍      | 205/600 [15:04<29:54,  4.54s/it]
 34%|███▍      | 206/600 [15:08<29:33,  4.50s/it]
 34%|███▍      | 207/600 [15:12<29:14,  4.47s/it]
 35%|███▍      | 208/600 [15:17<29:00,  4.44s/it]
 35%|███▍      | 209/600 [15:21<28:22,  4.36s/it]
 35%|███▌      | 210/600 [15:25<28:14,  4.35s/it]
 35%|███▌      | 211/600 [15:30<29:24,  4.54s/it]
 35%|███▌      | 212/600 [15:35<28:58,  4.48s/it]
 36%|███▌      | 213/600 [15:39<28:34,  4.43s/it]
 36%|███▌      | 214/600 [15:43<28:30,  4.43s/it]
 36%|███▌      | 215/600 [15:48<28:22,  4.42s/it]
 36%|███▌      | 216/600 [15:52<28:04,  4.39s/it]
 36%|███▌      | 217/600 [15:56<28:02,  4.39s/it]
 36%|███▋      | 218/600 [16:01<27:55,  4.39s/it]
 36%|███▋      | 219/600 [16:05<27:53,  4.39s/it]
 37%|███▋      | 220/600 [16:10<27:53,  4.40s/it]
                                                 
 37%|███▋      | 220/600 [16:10<27:53,  4.40s/it]
 37%|███▋      | 221/600 [16:14<27:48,  4.40s/it]
 37%|███▋      | 222/600 [16:18<27:47,  4.41s/it]
 37%|███▋      | 223/600 [16:23<27:43,  4.41s/it]
 37%|███▋      | 224/600 [16:27<27:44,  4.43s/it]
 38%|███▊      | 225/600 [16:32<27:36,  4.42s/it]
 38%|███▊      | 226/600 [16:36<28:13,  4.53s/it]
 38%|███▊      | 227/600 [16:41<27:41,  4.45s/it]
 38%|███▊      | 228/600 [16:45<27:33,  4.45s/it]
 38%|███▊      | 229/600 [16:49<27:09,  4.39s/it]
 38%|███▊      | 230/600 [16:54<27:01,  4.38s/it]
 38%|███▊      | 231/600 [16:58<26:56,  4.38s/it]
 39%|███▊      | 232/600 [17:02<26:41,  4.35s/it]
 39%|███▉      | 233/600 [17:07<26:44,  4.37s/it]
 39%|███▉      | 234/600 [17:11<26:21,  4.32s/it]
 39%|███▉      | 235/600 [17:16<26:29,  4.36s/it]
 39%|███▉      | 236/600 [17:20<26:31,  4.37s/it]
 40%|███▉      | 237/600 [17:24<26:21,  4.36s/it]
 40%|███▉      | 238/600 [17:28<25:58,  4.31s/it]
 40%|███▉      | 239/600 [17:33<26:58,  4.48s/it]
 40%|████      | 240/600 [17:38<26:29,  4.42s/it]
                                                 
 40%|████      | 240/600 [17:38<26:29,  4.42s/it]
 40%|████      | 241/600 [17:42<26:24,  4.41s/it]
 40%|████      | 242/600 [17:46<26:06,  4.38s/it]
 40%|████      | 243/600 [17:51<26:06,  4.39s/it]
 41%|████      | 244/600 [17:55<26:02,  4.39s/it]
 41%|████      | 245/600 [17:59<25:54,  4.38s/it]
 41%|████      | 246/600 [18:04<25:20,  4.30s/it]
 41%|████      | 247/600 [18:08<25:14,  4.29s/it]
 41%|████▏     | 248/600 [18:12<25:25,  4.33s/it]
 42%|████▏     | 249/600 [18:17<25:24,  4.34s/it]
 42%|████▏     | 250/600 [18:21<25:26,  4.36s/it]
 42%|████▏     | 251/600 [18:25<25:24,  4.37s/it]
 42%|████▏     | 252/600 [18:30<25:23,  4.38s/it]
 42%|████▏     | 253/600 [18:34<25:21,  4.39s/it]
 42%|████▏     | 254/600 [18:39<25:23,  4.40s/it]
 42%|████▎     | 255/600 [18:44<26:13,  4.56s/it]
 43%|████▎     | 256/600 [18:48<25:38,  4.47s/it]
 43%|████▎     | 257/600 [18:52<25:24,  4.45s/it]
 43%|████▎     | 258/600 [18:57<25:06,  4.40s/it]
 43%|████▎     | 259/600 [19:01<24:51,  4.38s/it]
 43%|████▎     | 260/600 [19:05<24:51,  4.39s/it]
                                                 
 43%|████▎     | 260/600 [19:05<24:51,  4.39s/it]
 44%|████▎     | 261/600 [19:10<24:48,  4.39s/it]
 44%|████▎     | 262/600 [19:14<24:38,  4.38s/it]
 44%|████▍     | 263/600 [19:18<24:07,  4.30s/it]
 44%|████▍     | 264/600 [19:23<24:17,  4.34s/it]
 44%|████▍     | 265/600 [19:27<24:15,  4.34s/it]
 44%|████▍     | 266/600 [19:31<23:40,  4.25s/it]
 44%|████▍     | 267/600 [19:35<23:38,  4.26s/it]
 45%|████▍     | 268/600 [19:40<23:44,  4.29s/it]
 45%|████▍     | 269/600 [19:44<23:49,  4.32s/it]
 45%|████▌     | 270/600 [19:48<23:50,  4.33s/it]
 45%|████▌     | 271/600 [19:53<24:41,  4.50s/it]
 45%|████▌     | 272/600 [19:58<24:26,  4.47s/it]
 46%|████▌     | 273/600 [20:02<24:17,  4.46s/it]
 46%|████▌     | 274/600 [20:06<24:05,  4.43s/it]
 46%|████▌     | 275/600 [20:11<23:56,  4.42s/it]
 46%|████▌     | 276/600 [20:15<23:49,  4.41s/it]
 46%|████▌     | 277/600 [20:19<23:21,  4.34s/it]
 46%|████▋     | 278/600 [20:24<23:14,  4.33s/it]
 46%|████▋     | 279/600 [20:28<23:15,  4.35s/it]
 47%|████▋     | 280/600 [20:32<23:10,  4.35s/it]
                                                 
 47%|████▋     | 280/600 [20:32<23:10,  4.35s/it]
 47%|████▋     | 281/600 [20:37<23:04,  4.34s/it]
 47%|████▋     | 282/600 [20:41<23:04,  4.35s/it]
 47%|████▋     | 283/600 [20:46<23:06,  4.37s/it]
 47%|████▋     | 284/600 [20:50<23:02,  4.37s/it]
 48%|████▊     | 285/600 [20:54<22:54,  4.36s/it]
 48%|████▊     | 286/600 [20:59<23:36,  4.51s/it]
 48%|████▊     | 287/600 [21:03<23:11,  4.44s/it]
 48%|████▊     | 288/600 [21:08<23:03,  4.43s/it]
 48%|████▊     | 289/600 [21:12<22:54,  4.42s/it]
 48%|████▊     | 290/600 [21:17<22:45,  4.41s/it]
 48%|████▊     | 291/600 [21:21<22:28,  4.36s/it]
 49%|████▊     | 292/600 [21:25<22:25,  4.37s/it]
 49%|████▉     | 293/600 [21:30<22:21,  4.37s/it]
 49%|████▉     | 294/600 [21:34<22:19,  4.38s/it]
 49%|████▉     | 295/600 [21:38<22:16,  4.38s/it]
 49%|████▉     | 296/600 [21:43<22:05,  4.36s/it]
 50%|████▉     | 297/600 [21:47<21:59,  4.36s/it]
 50%|████▉     | 298/600 [21:52<22:03,  4.38s/it]
 50%|████▉     | 299/600 [21:56<22:00,  4.39s/it]
 50%|█████     | 300/600 [22:01<22:29,  4.50s/it]
                                                 
 50%|█████     | 300/600 [22:01<22:29,  4.50s/it]
 50%|█████     | 301/600 [22:05<22:03,  4.43s/it]
 50%|█████     | 302/600 [22:09<21:47,  4.39s/it]
 50%|█████     | 303/600 [22:14<21:35,  4.36s/it]
 51%|█████     | 304/600 [22:18<21:19,  4.32s/it]
 51%|█████     | 305/600 [22:22<21:21,  4.34s/it]
 51%|█████     | 306/600 [22:26<21:11,  4.33s/it]
 51%|█████     | 307/600 [22:31<21:10,  4.33s/it]
 51%|█████▏    | 308/600 [22:35<21:00,  4.32s/it]
 52%|█████▏    | 309/600 [22:39<20:55,  4.32s/it]
 52%|█████▏    | 310/600 [22:44<20:50,  4.31s/it]
 52%|█████▏    | 311/600 [22:48<20:42,  4.30s/it]
 52%|█████▏    | 312/600 [22:52<20:38,  4.30s/it]
 52%|█████▏    | 313/600 [22:57<21:30,  4.50s/it]
 52%|█████▏    | 314/600 [23:02<21:08,  4.44s/it]
 52%|█████▎    | 315/600 [23:06<20:51,  4.39s/it]
 53%|█████▎    | 316/600 [23:10<20:48,  4.40s/it]
 53%|█████▎    | 317/600 [23:15<20:38,  4.38s/it]
 53%|█████▎    | 318/600 [23:19<20:34,  4.38s/it]
 53%|█████▎    | 319/600 [23:23<20:29,  4.37s/it]
 53%|█████▎    | 320/600 [23:28<20:14,  4.34s/it]
                                                 
 53%|█████▎    | 320/600 [23:28<20:14,  4.34s/it]
 54%|█████▎    | 321/600 [23:32<20:16,  4.36s/it]
 54%|█████▎    | 322/600 [23:36<20:12,  4.36s/it]
 54%|█████▍    | 323/600 [23:41<20:12,  4.38s/it]
 54%|█████▍    | 324/600 [23:45<20:09,  4.38s/it]
 54%|��████▍    | 325/600 [23:49<19:52,  4.34s/it]
 54%|█████▍    | 326/600 [23:54<19:51,  4.35s/it]
 55%|█████▍    | 327/600 [23:58<19:40,  4.33s/it]
 55%|█████▍    | 328/600 [24:02<19:43,  4.35s/it]
 55%|█████▍    | 329/600 [24:07<19:41,  4.36s/it]
 55%|█████▌    | 330/600 [24:12<20:28,  4.55s/it]
 55%|█████▌    | 331/600 [24:16<20:14,  4.52s/it]
 55%|█████▌    | 332/600 [24:21<20:00,  4.48s/it]
 56%|█████▌    | 333/600 [24:25<19:50,  4.46s/it]
 56%|█████▌    | 334/600 [24:29<19:38,  4.43s/it]
 56%|█████▌    | 335/600 [24:34<19:33,  4.43s/it]
 56%|█████▌    | 336/600 [24:38<19:10,  4.36s/it]
 56%|█████▌    | 337/600 [24:42<19:08,  4.37s/it]
 56%|█████▋    | 338/600 [24:47<19:09,  4.39s/it]
 56%|█████▋    | 339/600 [24:51<19:06,  4.39s/it]
 57%|█████▋    | 340/600 [24:56<19:03,  4.40s/it]
                                                 
 57%|█████▋    | 340/600 [24:56<19:03,  4.40s/it]
 57%|█████▋    | 341/600 [25:00<18:58,  4.40s/it]
 57%|█████▋    | 342/600 [25:04<18:40,  4.34s/it]
 57%|█████▋    | 343/600 [25:09<18:37,  4.35s/it]
 57%|█████▋    | 344/600 [25:13<18:26,  4.32s/it]
 57%|█████▊    | 345/600 [25:17<18:28,  4.35s/it]
 58%|█████▊    | 346/600 [25:22<19:09,  4.52s/it]
 58%|█████▊    | 347/600 [25:26<18:43,  4.44s/it]
 58%|█████▊    | 348/600 [25:31<18:15,  4.35s/it]
 58%|█████▊    | 349/600 [25:35<17:57,  4.29s/it]
 58%|█████▊    | 350/600 [25:39<17:28,  4.19s/it]
 58%|█████▊    | 351/600 [25:43<17:40,  4.26s/it]
 59%|█████▊    | 352/600 [25:47<17:42,  4.28s/it]
 59%|█████▉    | 353/600 [25:52<17:30,  4.25s/it]
 59%|█████▉    | 354/600 [25:56<17:35,  4.29s/it]
 59%|█████▉    | 355/600 [26:00<17:20,  4.25s/it]
 59%|█████▉    | 356/600 [26:05<17:27,  4.29s/it]
 60%|█████▉    | 357/600 [26:09<17:28,  4.32s/it]
 60%|█████▉    | 358/600 [26:13<17:32,  4.35s/it]
 60%|█████▉    | 359/600 [26:18<17:27,  4.35s/it]
 60%|██████    | 360/600 [26:22<17:25,  4.35s/it]
                                                 
 60%|██████    | 360/600 [26:22<17:25,  4.35s/it]
 60%|██████    | 361/600 [26:27<18:02,  4.53s/it]
 60%|██████    | 362/600 [26:31<17:37,  4.45s/it]
 60%|██████    | 363/600 [26:36<17:32,  4.44s/it]
 61%|██████    | 364/600 [26:40<17:24,  4.42s/it]
 61%|██████    | 365/600 [26:44<17:02,  4.35s/it]
 61%|██████    | 366/600 [26:49<16:48,  4.31s/it]
 61%|██████    | 367/600 [26:53<16:52,  4.34s/it]
 61%|██████▏   | 368/600 [26:57<16:35,  4.29s/it]
 62%|██████▏   | 369/600 [27:01<16:32,  4.30s/it]
 62%|██████▏   | 370/600 [27:06<16:32,  4.32s/it]
 62%|██████▏   | 371/600 [27:10<16:30,  4.33s/it]
 62%|██████▏   | 372/600 [27:15<16:34,  4.36s/it]
 62%|██████▏   | 373/600 [27:19<16:31,  4.37s/it]
 62%|██████▏   | 374/600 [27:24<17:03,  4.53s/it]
 62%|██████▎   | 375/600 [27:28<16:40,  4.45s/it]
 63%|██████▎   | 376/600 [27:33<16:32,  4.43s/it]
 63%|██████▎   | 377/600 [27:37<16:20,  4.40s/it]
 63%|██████▎   | 378/600 [27:41<16:15,  4.39s/it]
 63%|██████▎   | 379/600 [27:46<16:12,  4.40s/it]
 63%|██████▎   | 380/600 [27:50<16:06,  4.39s/it]
                                                 
 63%|██████▎   | 380/600 [27:50<16:06,  4.39s/it]
 64%|██████▎   | 381/600 [27:54<16:04,  4.40s/it]
 64%|██████▎   | 382/600 [27:59<15:48,  4.35s/it]
 64%|██████▍   | 383/600 [28:03<15:35,  4.31s/it]
 64%|██████▍   | 384/600 [28:07<15:32,  4.32s/it]
 64%|██████▍   | 385/600 [28:12<15:33,  4.34s/it]
 64%|██████▍   | 386/600 [28:16<15:31,  4.35s/it]
 64%|██████▍   | 387/600 [28:20<15:27,  4.35s/it]
 65%|██████▍   | 388/600 [28:25<15:21,  4.34s/it]
 65%|██████▍   | 389/600 [28:29<15:18,  4.35s/it]
 65%|██████▌   | 390/600 [28:33<15:16,  4.37s/it]
 65%|██████▌   | 391/600 [28:38<15:52,  4.56s/it]
 65%|██████▌   | 392/600 [28:43<15:41,  4.52s/it]
 66%|██████▌   | 393/600 [28:47<15:29,  4.49s/it]
 66%|██████▌   | 394/600 [28:52<15:21,  4.47s/it]
 66%|██████▌   | 395/600 [28:56<15:13,  4.46s/it]
 66%|██████▌   | 396/600 [29:00<15:01,  4.42s/it]
 66%|██████▌   | 397/600 [29:05<14:52,  4.40s/it]
 66%|██████▋   | 398/600 [29:09<14:50,  4.41s/it]
 66%|██████▋   | 399/600 [29:14<14:46,  4.41s/it]
 67%|██████▋   | 400/600 [29:18<14:29,  4.35s/it]
                                                 
 67%|██████▋   | 400/600 [29:18<14:29,  4.35s/it]/usr/local/lib/python3.10/dist-packages/peft/utils/save_and_load.py:295: UserWarning: Could not find a config file in Qwen/Qwen3-4B-Base - will assume that the vocabulary was not modified.
+  warnings.warn(
+
 67%|██████▋   | 401/600 [29:25<16:50,  5.08s/it]
 67%|██████▋   | 402/600 [29:29<16:04,  4.87s/it]
 67%|██████▋   | 403/600 [29:33<15:28,  4.71s/it]
 67%|██████▋   | 404/600 [29:38<14:59,  4.59s/it]
 68%|██████▊   | 405/600 [29:42<14:30,  4.47s/it]
 68%|██████▊   | 406/600 [29:47<14:55,  4.62s/it]
 68%|██████▊   | 407/600 [29:51<14:39,  4.56s/it]
 68%|██████▊   | 408/600 [29:56<14:19,  4.48s/it]
 68%|██████▊   | 409/600 [30:00<14:02,  4.41s/it]
 68%|██████▊   | 410/600 [30:04<13:57,  4.41s/it]
 68%|██████▊   | 411/600 [30:08<13:45,  4.37s/it]
 69%|██████▊   | 412/600 [30:13<13:32,  4.32s/it]
 69%|██████▉   | 413/600 [30:17<13:33,  4.35s/it]
 69%|██████▉   | 414/600 [30:22<13:31,  4.36s/it]
 69%|██████▉   | 415/600 [30:26<13:15,  4.30s/it]
 69%|██████▉   | 416/600 [30:30<13:14,  4.32s/it]
 70%|██████▉   | 417/600 [30:34<13:14,  4.34s/it]
 70%|██████▉   | 418/600 [30:39<13:05,  4.32s/it]
 70%|██████▉   | 419/600 [30:43<12:51,  4.26s/it]
 70%|███████   | 420/600 [30:47<12:48,  4.27s/it]
                                                 
 70%|███████   | 420/600 [30:47<12:48,  4.27s/it]
 70%|███████   | 421/600 [30:52<12:51,  4.31s/it]
 70%|███████   | 422/600 [30:56<13:18,  4.49s/it]
 70%|███████   | 423/600 [31:01<13:11,  4.47s/it]
 71%|███████   | 424/600 [31:05<13:03,  4.45s/it]
 71%|███████   | 425/600 [31:10<12:56,  4.44s/it]
 71%|███████   | 426/600 [31:14<12:41,  4.38s/it]
 71%|███████   | 427/600 [31:18<12:35,  4.37s/it]
 71%|███████▏  | 428/600 [31:23<12:29,  4.36s/it]
 72%|███████▏  | 429/600 [31:27<12:27,  4.37s/it]
 72%|███████▏  | 430/600 [31:31<12:25,  4.39s/it]
 72%|███████▏  | 431/600 [31:36<12:14,  4.35s/it]
 72%|███████▏  | 432/600 [31:40<12:11,  4.35s/it]
 72%|███████▏  | 433/600 [31:44<12:11,  4.38s/it]
 72%|███████▏  | 434/600 [31:49<12:07,  4.38s/it]
 72%|███████▎  | 435/600 [31:53<12:03,  4.38s/it]
 73%|███████▎  | 436/600 [31:57<11:49,  4.33s/it]
 73%|███████▎  | 437/600 [32:02<11:49,  4.36s/it]
 73%|███████▎  | 438/600 [32:07<12:15,  4.54s/it]
 73%|███████▎  | 439/600 [32:11<12:05,  4.50s/it]
 73%|███████▎  | 440/600 [32:16<11:55,  4.47s/it]
                                                 
 73%|███████▎  | 440/600 [32:16<11:55,  4.47s/it]
 74%|███████▎  | 441/600 [32:20<11:43,  4.43s/it]
 74%|███████▎  | 442/600 [32:24<11:37,  4.41s/it]
 74%|███████▍  | 443/600 [32:29<11:31,  4.40s/it]
 74%|███████▍  | 444/600 [32:33<11:27,  4.41s/it]
 74%|███████▍  | 445/600 [32:38<11:22,  4.40s/it]
 74%|███████▍  | 446/600 [32:42<11:16,  4.40s/it]
 74%|███████▍  | 447/600 [32:46<11:06,  4.35s/it]
 75%|███████▍  | 448/600 [32:51<11:03,  4.36s/it]
 75%|███████▍  | 449/600 [32:55<10:59,  4.37s/it]
 75%|███████▌  | 450/600 [32:59<10:57,  4.38s/it]
 75%|███████▌  | 451/600 [33:04<10:49,  4.36s/it]
 75%|███████▌  | 452/600 [33:08<10:46,  4.37s/it]
 76%|███████▌  | 453/600 [33:13<10:58,  4.48s/it]
 76%|███████▌  | 454/600 [33:17<10:51,  4.46s/it]
 76%|███████▌  | 455/600 [33:22<10:41,  4.42s/it]
 76%|███████▌  | 456/600 [33:26<10:25,  4.35s/it]
 76%|███████▌  | 457/600 [33:30<10:14,  4.30s/it]
 76%|███████▋  | 458/600 [33:34<10:13,  4.32s/it]
 76%|███████▋  | 459/600 [33:39<10:12,  4.34s/it]
 77%|███████▋  | 460/600 [33:43<10:11,  4.37s/it]
                                                 
 77%|███████▋  | 460/600 [33:43<10:11,  4.37s/it]
 77%|███████▋  | 461/600 [33:47<10:09,  4.39s/it]
 77%|███████▋  | 462/600 [33:52<10:06,  4.39s/it]
 77%|███████▋  | 463/600 [33:56<10:02,  4.40s/it]
 77%|███████▋  | 464/600 [34:01<09:58,  4.40s/it]
 78%|███████▊  | 465/600 [34:05<09:55,  4.41s/it]
 78%|███████▊  | 466/600 [34:10<10:14,  4.59s/it]
 78%|███████▊  | 467/600 [34:14<09:48,  4.43s/it]
 78%|███████▊  | 468/600 [34:19<09:40,  4.40s/it]
 78%|█████��█▊  | 469/600 [34:23<09:36,  4.40s/it]
 78%|███████▊  | 470/600 [34:27<09:27,  4.36s/it]
 78%|███████▊  | 471/600 [34:32<09:25,  4.38s/it]
 79%|███████▊  | 472/600 [34:36<09:22,  4.39s/it]
 79%|███████▉  | 473/600 [34:40<09:15,  4.37s/it]
 79%|███████▉  | 474/600 [34:45<09:13,  4.39s/it]
 79%|███████▉  | 475/600 [34:49<08:54,  4.28s/it]
 79%|███████▉  | 476/600 [34:53<08:43,  4.22s/it]
 80%|███████▉  | 477/600 [34:57<08:39,  4.22s/it]
 80%|███████▉  | 478/600 [35:01<08:39,  4.26s/it]
 80%|███████▉  | 479/600 [35:06<08:42,  4.32s/it]
 80%|████████  | 480/600 [35:10<08:41,  4.35s/it]
                                                 
 80%|████████  | 480/600 [35:10<08:41,  4.35s/it]
 80%|████████  | 481/600 [35:15<08:37,  4.35s/it]
 80%|████████  | 482/600 [35:19<08:35,  4.37s/it]
 80%|████████  | 483/600 [35:24<08:52,  4.55s/it]
 81%|████████  | 484/600 [35:29<08:43,  4.51s/it]
 81%|████████  | 485/600 [35:33<08:33,  4.47s/it]
 81%|████████  | 486/600 [35:37<08:26,  4.44s/it]
 81%|████████  | 487/600 [35:42<08:20,  4.43s/it]
 81%|████████▏ | 488/600 [35:46<08:14,  4.42s/it]
 82%|████████▏ | 489/600 [35:50<08:09,  4.41s/it]
 82%|████████▏ | 490/600 [35:55<08:01,  4.37s/it]
 82%|████████▏ | 491/600 [35:59<07:57,  4.38s/it]
 82%|████████▏ | 492/600 [36:03<07:50,  4.35s/it]
 82%|████████▏ | 493/600 [36:08<07:46,  4.36s/it]
 82%|████████▏ | 494/600 [36:12<07:42,  4.36s/it]
 82%|████████▎ | 495/600 [36:17<07:39,  4.38s/it]
 83%|████████▎ | 496/600 [36:21<07:33,  4.36s/it]
 83%|████████▎ | 497/600 [36:25<07:29,  4.36s/it]
 83%|████████▎ | 498/600 [36:30<07:26,  4.38s/it]
 83%|████████▎ | 499/600 [36:35<07:39,  4.55s/it]
 83%|████████▎ | 500/600 [36:39<07:31,  4.51s/it]
                                                 
 83%|████████▎ | 500/600 [36:39<07:31,  4.51s/it]
 84%|████████▎ | 501/600 [36:43<07:24,  4.49s/it]
 84%|████████▎ | 502/600 [36:48<07:17,  4.47s/it]
 84%|████████▍ | 503/600 [36:52<07:08,  4.42s/it]
 84%|████████▍ | 504/600 [36:57<07:04,  4.42s/it]
 84%|████████▍ | 505/600 [37:01<06:59,  4.42s/it]
 84%|████████▍ | 506/600 [37:05<06:52,  4.38s/it]
 84%|████████▍ | 507/600 [37:10<06:46,  4.37s/it]
 85%|████████▍ | 508/600 [37:14<06:40,  4.35s/it]
 85%|████████▍ | 509/600 [37:18<06:33,  4.33s/it]
 85%|████████▌ | 510/600 [37:23<06:29,  4.33s/it]
 85%|████████▌ | 511/600 [37:27<06:26,  4.35s/it]
 85%|████████▌ | 512/600 [37:31<06:24,  4.37s/it]
 86%|████████▌ | 513/600 [37:36<06:21,  4.38s/it]
 86%|████████▌ | 514/600 [37:41<06:27,  4.51s/it]
 86%|████████▌ | 515/600 [37:45<06:19,  4.47s/it]
 86%|████████▌ | 516/600 [37:49<06:13,  4.45s/it]
 86%|████████▌ | 517/600 [37:54<06:04,  4.40s/it]
 86%|████████▋ | 518/600 [37:58<06:01,  4.41s/it]
 86%|████████▋ | 519/600 [38:03<05:57,  4.41s/it]
 87%|████████▋ | 520/600 [38:07<05:52,  4.41s/it]
                                                 
 87%|████████▋ | 520/600 [38:07<05:52,  4.41s/it]
 87%|████████▋ | 521/600 [38:11<05:42,  4.34s/it]
 87%|████████▋ | 522/600 [38:15<05:38,  4.34s/it]
 87%|████████▋ | 523/600 [38:20<05:30,  4.29s/it]
 87%|████████▋ | 524/600 [38:24<05:28,  4.32s/it]
 88%|████████▊ | 525/600 [38:28<05:26,  4.36s/it]
 88%|████████▊ | 526/600 [38:33<05:21,  4.35s/it]
 88%|████████▊ | 527/600 [38:37<05:10,  4.26s/it]
 88%|████████▊ | 528/600 [38:42<05:22,  4.48s/it]
 88%|████████▊ | 529/600 [38:46<05:16,  4.46s/it]
 88%|████████▊ | 530/600 [38:51<05:10,  4.44s/it]
 88%|████████▊ | 531/600 [38:55<05:05,  4.43s/it]
 89%|████████▊ | 532/600 [38:59<04:56,  4.36s/it]
 89%|████████▉ | 533/600 [39:04<04:52,  4.37s/it]
 89%|████████▉ | 534/600 [39:08<04:48,  4.37s/it]
 89%|████████▉ | 535/600 [39:12<04:44,  4.37s/it]
 89%|████████▉ | 536/600 [39:17<04:40,  4.39s/it]
 90%|████████▉ | 537/600 [39:21<04:36,  4.38s/it]
 90%|████████▉ | 538/600 [39:25<04:30,  4.36s/it]
 90%|████████▉ | 539/600 [39:30<04:27,  4.38s/it]
 90%|█████████ | 540/600 [39:34<04:20,  4.34s/it]
                                                 
 90%|█████████ | 540/600 [39:34<04:20,  4.34s/it]
 90%|█████████ | 541/600 [39:39<04:17,  4.36s/it]
 90%|█████████ | 542/600 [39:43<04:13,  4.36s/it]
 90%|█████████ | 543/600 [39:48<04:19,  4.56s/it]
 91%|█████████ | 544/600 [39:52<04:12,  4.51s/it]
 91%|█████████ | 545/600 [39:57<04:05,  4.46s/it]
 91%|█████████ | 546/600 [40:01<03:58,  4.42s/it]
 91%|█████████ | 547/600 [40:05<03:52,  4.39s/it]
 91%|█████████▏| 548/600 [40:10<03:47,  4.37s/it]
 92%|█████████▏| 549/600 [40:14<03:43,  4.37s/it]
 92%|█████████▏| 550/600 [40:19<03:39,  4.39s/it]
 92%|█████████▏| 551/600 [40:23<03:32,  4.34s/it]
 92%|█████████▏| 552/600 [40:27<03:28,  4.34s/it]
 92%|█████████▏| 553/600 [40:31<03:24,  4.36s/it]
 92%|█████████▏| 554/600 [40:36<03:20,  4.36s/it]
 92%|█████████▎| 555/600 [40:40<03:17,  4.38s/it]
 93%|█████████▎| 556/600 [40:45<03:12,  4.38s/it]
 93%|█████████▎| 557/600 [40:49<03:08,  4.39s/it]
 93%|█████████▎| 558/600 [40:53<03:03,  4.36s/it]
 93%|█████████▎| 559/600 [40:58<02:59,  4.37s/it]
 93%|█████████▎| 560/600 [41:03<03:02,  4.55s/it]
                                                 
 93%|█████████▎| 560/600 [41:03<03:02,  4.55s/it]
 94%|█████████▎| 561/600 [41:07<02:54,  4.47s/it]
 94%|█████████▎| 562/600 [41:11<02:49,  4.46s/it]
 94%|█████████▍| 563/600 [41:16<02:44,  4.44s/it]
 94%|█████████▍| 564/600 [41:20<02:38,  4.42s/it]
 94%|█████████▍| 565/600 [41:25<02:34,  4.42s/it]
 94%|█████████▍| 566/600 [41:29<02:30,  4.42s/it]
 94%|█████████▍| 567/600 [41:33<02:25,  4.41s/it]
 95%|█████████▍| 568/600 [41:38<02:21,  4.42s/it]
 95%|█████████▍| 569/600 [41:42<02:15,  4.38s/it]
 95%|█████████▌| 570/600 [41:47<02:11,  4.40s/it]
 95%|█████████▌| 571/600 [41:51<02:07,  4.39s/it]
 95%|█████████▌| 572/600 [41:55<02:03,  4.40s/it]
 96%|█████████▌| 573/600 [42:00<01:57,  4.34s/it]
 96%|█████████▌| 574/600 [42:04<01:53,  4.35s/it]
 96%|█████████▌| 575/600 [42:09<01:52,  4.50s/it]
 96%|█████████▌| 576/600 [42:13<01:47,  4.46s/it]
 96%|█████████▌| 577/600 [42:18<01:41,  4.43s/it]
 96%|█████████▋| 578/600 [42:22<01:36,  4.39s/it]
 96%|█████████▋| 579/600 [42:26<01:32,  4.39s/it]
 97%|█████████▋| 580/600 [42:31<01:27,  4.38s/it]
                                                 
 97%|█████████▋| 580/600 [42:31<01:27,  4.38s/it]
 97%|█████████▋| 581/600 [42:35<01:23,  4.39s/it]
 97%|█████████▋| 582/600 [42:39<01:18,  4.34s/it]
 97%|█████████▋| 583/600 [42:44<01:14,  4.37s/it]
 97%|█████████▋| 584/600 [42:48<01:10,  4.38s/it]
 98%|█████████▊| 585/600 [42:52<01:05,  4.35s/it]
 98%|█████████▊| 586/600 [42:57<01:00,  4.34s/it]
 98%|█████████▊| 587/600 [43:01<00:56,  4.37s/it]
 98%|█████████▊| 588/600 [43:05<00:52,  4.37s/it]
 98%|█████████▊| 589/600 [43:10<00:50,  4.57s/it]
 98%|█████████▊| 590/600 [43:15<00:44,  4.50s/it]
 98%|█████████▊| 591/600 [43:19<00:40,  4.48s/it]
 99%|█████████▊| 592/600 [43:24<00:35,  4.44s/it]
 99%|█████████▉| 593/600 [43:28<00:30,  4.43s/it]
 99%|█████████▉| 594/600 [43:32<00:26,  4.42s/it]
 99%|█████████▉| 595/600 [43:37<00:21,  4.39s/it]
 99%|█████████▉| 596/600 [43:41<00:17,  4.38s/it]
                                                 
+  warnings.warn(
+
                                                 
+{'loss': 0.2335, 'grad_norm': 0.23258326947689056, 'learning_rate': 0.00019999854312354064, 'epoch': 0.01}
+{'loss': 0.2345, 'grad_norm': 0.20497269928455353, 'learning_rate': 0.00019935820360309777, 'epoch': 0.02}
+{'loss': 0.231, 'grad_norm': 0.1947828233242035, 'learning_rate': 0.0001975609645454704, 'epoch': 0.03}
+{'loss': 0.2313, 'grad_norm': 0.190537229180336, 'learning_rate': 0.0001946277525068821, 'epoch': 0.04}
+{'loss': 0.2031, 'grad_norm': 0.17262256145477295, 'learning_rate': 0.0001905927209998447, 'epoch': 0.05}
+{'loss': 0.2179, 'grad_norm': 0.20835602283477783, 'learning_rate': 0.000185502852819072, 'epoch': 0.07}
+{'loss': 0.2139, 'grad_norm': 0.18527470529079437, 'learning_rate': 0.00017941741298676774, 'epoch': 0.08}
+{'loss': 0.2205, 'grad_norm': 0.18617987632751465, 'learning_rate': 0.00017240725868704218, 'epoch': 0.09}
+{'loss': 0.2221, 'grad_norm': 0.20183837413787842, 'learning_rate': 0.00016455401422438986, 'epoch': 0.1}
+{'loss': 0.2174, 'grad_norm': 0.19644296169281006, 'learning_rate': 0.00015594912061278626, 'epoch': 0.11}
+{'loss': 0.2161, 'grad_norm': 0.19903691112995148, 'learning_rate': 0.00014669277086172406, 'epoch': 0.12}
+{'loss': 0.2049, 'grad_norm': 0.19178642332553864, 'learning_rate': 0.00013689274335642652, 'epoch': 0.13}
+{'loss': 0.215, 'grad_norm': 0.19386041164398193, 'learning_rate': 0.00012666314691603617, 'epoch': 0.14}
+{'loss': 0.2022, 'grad_norm': 0.14429228007793427, 'learning_rate': 0.00011612309214197599, 'epoch': 0.15}
+{'loss': 0.2118, 'grad_norm': 0.19819147884845734, 'learning_rate': 0.00010539530452693625, 'epoch': 0.16}
+{'loss': 0.2068, 'grad_norm': 0.16852140426635742, 'learning_rate': 9.460469547306376e-05, 'epoch': 0.18}
+{'loss': 0.2222, 'grad_norm': 0.17052850127220154, 'learning_rate': 8.387690785802402e-05, 'epoch': 0.19}
+{'loss': 0.2059, 'grad_norm': 0.1543218344449997, 'learning_rate': 7.333685308396383e-05, 'epoch': 0.2}
+{'loss': 0.2039, 'grad_norm': 0.1840507835149765, 'learning_rate': 6.310725664357349e-05, 'epoch': 0.21}
+{'loss': 0.2195, 'grad_norm': 0.15787966549396515, 'learning_rate': 5.3307229138275936e-05, 'epoch': 0.22}
+{'loss': 0.2035, 'grad_norm': 0.23087719082832336, 'learning_rate': 4.4050879387213764e-05, 'epoch': 0.23}
+{'loss': 0.2104, 'grad_norm': 0.19456110894680023, 'learning_rate': 3.544598577561016e-05, 'epoch': 0.24}
+{'loss': 0.2036, 'grad_norm': 0.14935997128486633, 'learning_rate': 2.759274131295787e-05, 'epoch': 0.25}
+{'loss': 0.22, 'grad_norm': 0.169290229678154, 'learning_rate': 2.0582587013232267e-05, 'epoch': 0.26}
+{'loss': 0.2343, 'grad_norm': 0.17034634947776794, 'learning_rate': 1.4497147180928027e-05, 'epoch': 0.27}
+{'loss': 0.2005, 'grad_norm': 0.21966595947742462, 'learning_rate': 9.407279000155312e-06, 'epoch': 0.28}
+{'loss': 0.2022, 'grad_norm': 0.18288274109363556, 'learning_rate': 5.372247493117922e-06, 'epoch': 0.3}
+{'loss': 0.2009, 'grad_norm': 0.1908612698316574, 'learning_rate': 2.4390354545296255e-06, 'epoch': 0.31}
+{'loss': 0.1993, 'grad_norm': 0.15066727995872498, 'learning_rate': 6.41796396902239e-07, 'epoch': 0.32}
+{'loss': 0.2067, 'grad_norm': 0.16710063815116882, 'learning_rate': 1.4568764593603234e-09, 'epoch': 0.33}
+{'train_runtime': 2641.7884, 'train_samples_per_second': 3.634, 'train_steps_per_second': 0.227, 'train_loss': 0.21382807016372682, 'epoch': 0.33}
+train time (s): 2642.660863876343
+/usr/local/lib/python3.10/dist-packages/peft/utils/save_and_load.py:295: UserWarning: Could not find a config file in Qwen/Qwen3-4B-Base - will assume that the vocabulary was not modified.
+  warnings.warn(
+saved to out/main1

instances/end_of_run/d05/state/probe_eval.json

@@ -0,0 +1,4 @@
+{
+  "accuracy": 0.0,
+  "stderr": 0.0
+}

instances/end_of_run/d05/state/scripts/build_data.py

@@ -0,0 +1,130 @@
+import os, re, ast, json, glob, random, sys
+import pandas as pd
+
+random.seed(0)
+INSTRUCTION = "\nRead the following function signature and docstring, and fully implement\nthe function described. Your response should only contain the code for\nthis function.\n"
+
+def find_def_block(code):
+    """Return (signature_lines, docstring_lines) of the first module-level function def."""
+    lines = code.split("\n")
+    start = None
+    for i, ln in enumerate(lines):
+        if re.match(r'^def\s+\w+\s*\(', ln):
+            start = i; break
+        if ln.strip() and not ln.startswith((' ', '\t')) and not ln.startswith(('#','@','"""',"'''")):
+            # module-level non-def statement before def
+            pass
+    if start is None:
+        return None
+    # find end of signature (balanced parens ending with ':')
+    depth = 0; j = start; sig_end = None
+    while j < len(lines):
+        depth += lines[j].count('(') - lines[j].count(')')
+        if depth <= 0 and lines[j].rstrip().endswith(':'):
+            sig_end = j; break
+        j += 1
+    if sig_end is None:
+        return None
+    sig = lines[start:sig_end+1]
+    # docstring
+    k = sig_end + 1
+    doc = []
+    if k < len(lines) and lines[k].strip().startswith(('"""', "'''")):
+        q = '"""' if lines[k].strip().startswith('"""') else "'''"
+        doc.append(lines[k])
+        if lines[k].strip().count(q) < 2:
+            k += 1
+            while k < len(lines):
+                doc.append(lines[k])
+                if q in lines[k]:
+                    break
+                k += 1
+    return sig, doc
+
+def clean_code(code):
+    if not isinstance(code, str): return None
+    code = code.replace("\r\n", "\n").replace("\t", "    ")
+    if "```" in code:
+        m = re.search(r"```(?:python)?\n(.*?)```", code, re.DOTALL)
+        if m: code = m.group(1)
+        else: return None
+    return code.strip("\n")
+
+def make_example(context_prose, code):
+    code = clean_code(code)
+    if not code: return None
+    try:
+        ast.parse(code)
+    except Exception:
+        return None
+    db = find_def_block(code)
+    if db is None: return None
+    sig, doc = db
+    block = "\n".join(sig + doc)
+    parts = []
+    if context_prose and context_prose.strip():
+        parts.append(context_prose.strip())
+    parts.append(block)
+    prompt_ctx = "\n\n".join(parts)
+    return {"prompt": INSTRUCTION + prompt_ctx + "\n\n", "completion": code}
+
+examples = []
+
+# ---- KodCode-V1 (Leetcode python) ----
+fs = sorted(glob.glob("/mnt/ben/hf_cache/hub/datasets--KodCode--KodCode-V1/snapshots/*/data/train-*.parquet"))
+kc = 0
+for f in fs:
+    df = pd.read_parquet(f, columns=["subset","question","solution"])
+    df = df[df["subset"] == "Leetcode"]
+    df = df.sample(frac=1.0, random_state=0)
+    for q, s in zip(df["question"].tolist(), df["solution"].tolist()):
+        e = make_example(q, s)
+        if e: examples.append(e); kc += 1
+        if kc >= 20000: break
+    if kc >= 20000: break
+print("kodcode:", kc, file=sys.stderr)
+
+# ---- MBPP (google-research) ----
+import datasets
+for split in ["train","test","validation"]:
+    try:
+        ds = datasets.load_dataset("google-research-datasets/mbpp", split=split)
+        for r in ds:
+            e = make_example(r.get("text"), r.get("code"))
+            if e: examples.append(e)
+    except Exception as ex:
+        print("mbpp err", split, ex, file=sys.stderr)
+
+# ---- Magicoder python ----
+try:
+    p = glob.glob("/mnt/ben/hf_cache/hub/datasets--ise-uiuc--Magicoder-OSS-Instruct-75K/snapshots/*/data-oss_instruct-decontaminated.jsonl")[0]
+    mg = 0
+    with open(p) as fh:
+        for line in fh:
+            r = json.loads(line)
+            if r.get("lang") != "python": continue
+            e = make_example(r.get("problem"), r.get("solution"))
+            if e: examples.append(e); mg += 1
+    print("magicoder py:", mg, file=sys.stderr)
+except Exception as ex:
+    print("magicoder err", ex, file=sys.stderr)
+
+# ---- self-oss-instruct-sc2 ----
+try:
+    f = glob.glob("/mnt/ben/hf_cache/hub/datasets--bigcode--self-oss-instruct-sc2-exec-filter-50k/snapshots/*/data/train-*.parquet")[0]
+    df = pd.read_parquet(f, columns=["instruction","response"])
+    df = df.sample(frac=1.0, random_state=0).head(20000)
+    so = 0
+    for ins, resp in zip(df["instruction"].tolist(), df["response"].tolist()):
+        e = make_example(ins, resp)
+        if e: examples.append(e); so += 1
+    print("self-oss:", so, file=sys.stderr)
+except Exception as ex:
+    print("self-oss err", ex, file=sys.stderr)
+
+random.shuffle(examples)
+os.makedirs("data", exist_ok=True)
+with open("data/train_all.jsonl","w") as fh:
+    for e in examples:
+        fh.write(json.dumps(e)+"\n")
+print("TOTAL:", len(examples))

instances/end_of_run/d05/state/scripts/check_detail.py

@@ -0,0 +1,30 @@
+import json, re, sys
+def norm(s):
+    s=s.lower(); s=re.sub(r'[^a-z0-9_]+',' ',s); return re.sub(r'\s+',' ',s).strip()
+def sh(text,n):
+    t=text.split(); return set(tuple(t[i:i+n]) for i in range(max(0,len(t)-n+1)))
+ref=json.load(open('/mnt/ben/test_data.json'))
+R=[]
+for i,r in enumerate(ref):
+    for fld in ('question','answer'):
+        R.append((i,fld,sh(norm(r[fld]),12)))
+rows=[]
+with open('data/train_all.jsonl') as f:
+    for li,line in enumerate(f):
+        line=line.strip()
+        if not line: continue
+        obj=json.loads(line); text=obj.get('prompt','')+" "+obj.get('completion','')
+        tn=norm(text); ts12=sh(tn,12)
+        best=0; bidx=None
+        for i,fld,s in R:
+            c=len(ts12&s)
+            if c>best: best=c; bidx=(i,fld)
+        if best>=6: rows.append((best,li,bidx,obj))
+rows.sort(reverse=True, key=lambda x:x[0])
+print("docs with >=6 shared 12-grams:", len(rows))
+for best,li,bidx,obj in rows[:8]:
+    print("="*70)
+    print("hits",best,"line",li,"ref",bidx)
+    print("PROMPT:", obj['prompt'][:300].replace("\n","\\n"))
+    print("COMPL:", obj['completion'][:400].replace("\n","\\n"))
+    if bidx: print("REFQ:", ref[bidx[0]]['question'][:250].replace("\n","\\n"))

instances/end_of_run/d05/state/scripts/filter_contam.py

@@ -0,0 +1,32 @@
+import json, re
+def norm(s):
+    s=s.lower(); s=re.sub(r'[^a-z0-9_]+',' ',s); return re.sub(r'\s+',' ',s).strip()
+def sh(text,n):
+    t=text.split(); return set(tuple(t[i:i+n]) for i in range(max(0,len(t)-n+1)))
+ref=json.load(open('/mnt/ben/test_data.json'))
+R=[]
+for i,r in enumerate(ref):
+    for fld in ('question','answer'):
+        R.append((i,fld,norm(r[fld]),sh(norm(r[fld]),12),sh(norm(r[fld]),25)))
+
+THRESH=3
+kept=0; dropped=[]
+with open('data/train_all.jsonl') as f, open('data/train_clean.jsonl','w') as out:
+    for li,line in enumerate(f):
+        line=line.strip()
+        if not line: continue
+        obj=json.loads(line)
+        text=obj.get('prompt','')+" \n "+obj.get('completion','')
+        tn=norm(text); ts12=sh(tn,12); ts25=sh(tn,25)
+        worst=0; wref=None
+        for i,fld,rtxt,s12,s25 in R:
+            c12=len(ts12&s12); c25=len(ts25&s25)
+            sc=c12+5*c25
+            if sc>worst: worst=sc; wref=(i,fld,c12,c25)
+        if worst>=THRESH:
+            dropped.append((worst,li,wref,obj))
+        else:
+            out.write(json.dumps(obj)+"\n"); kept+=1
+print("kept:",kept,"dropped:",len(dropped))
+# verify no long verbatim in kept
+import subprocess

instances/end_of_run/d05/state/scripts/gen_test.py

@@ -0,0 +1,18 @@
+import sys, torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+path=sys.argv[1]
+tok=AutoTokenizer.from_pretrained(path,trust_remote_code=True)
+m=AutoModelForCausalLM.from_pretrained(path,dtype=torch.bfloat16,device_map="cuda",trust_remote_code=True)
+m.eval()
+import json
+q=json.load(open('/mnt/ben/test_data.json'))
+INSTR="\nRead the following function signature and docstring, and fully implement\nthe function described. Your response should only contain the code for\nthis function.\n"
+for r in q[:3]:
+    prompt=f"<|im_start|>user\n{INSTR}{r['question']}<|im_end|>\n<|im_start|>assistant\n"
+    ids=tok(prompt,return_tensors="pt").to("cuda")
+    out=m.generate(**ids,max_new_tokens=600,do_sample=False,pad_token_id=tok.eos_token_id)
+    gen=out[0][ids['input_ids'].shape[1]:]
+    txt=tok.decode(gen)
+    print("="*60)
+    print("ENDS_WITH_IMEND:", txt.rstrip().endswith("<|im_end|>"))
+    print(repr(txt[-160:]))

instances/end_of_run/d05/state/scripts/gen_test2.py

@@ -0,0 +1,13 @@
+import sys, torch, json
+from transformers import AutoModelForCausalLM, AutoTokenizer
+path=sys.argv[1]
+tok=AutoTokenizer.from_pretrained(path,trust_remote_code=True)
+m=AutoModelForCausalLM.from_pretrained(path,dtype=torch.bfloat16,device_map="cuda",trust_remote_code=True); m.eval()
+rows=[json.loads(l) for i,l in enumerate(open("data/train_clean.jsonl")) if i<500]
+for r in rows[:4]:
+    prompt=f"<|im_start|>user\n{r['prompt']}<|im_end|>\n<|im_start|>assistant\n"
+    ids=tok(prompt,return_tensors="pt").to("cuda")
+    out=m.generate(**ids,max_new_tokens=400,do_sample=False,pad_token_id=tok.eos_token_id)
+    txt=tok.decode(out[0][ids['input_ids'].shape[1]:])
+    print("="*50, "ENDS_IMEND:", txt.rstrip().endswith("<|im_end|>"))
+    print(repr(txt[-140:]))

instances/end_of_run/d05/state/scripts/merge.py

@@ -0,0 +1,10 @@
+import sys, torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+base="Qwen/Qwen3-4B-Base"; adapter=sys.argv[1]; out=sys.argv[2]
+m=AutoModelForCausalLM.from_pretrained(base, torch_dtype=torch.bfloat16, trust_remote_code=True)
+m=PeftModel.from_pretrained(m, adapter)
+m=m.merge_and_unload()
+m.save_pretrained(out, safe_serialization=True)
+AutoTokenizer.from_pretrained(adapter).save_pretrained(out)
+print("merged ->", out)

instances/end_of_run/d05/state/scripts/my_contam_check.py

@@ -0,0 +1,58 @@
+import json, re, sys, argparse
+from collections import defaultdict
+
+def norm(s):
+    s = s.lower()
+    s = re.sub(r'[^a-z0-9_]+', ' ', s)
+    return re.sub(r'\s+', ' ', s).strip()
+
+def shingles(text, n):
+    toks = text.split()
+    if len(toks) < n: 
+        return set(), toks
+    return set(tuple(toks[i:i+n]) for i in range(len(toks)-n+1)), toks
+
+ref = json.load(open('/mnt/ben/test_data.json'))
+# Build reference n-gram sets at multiple sizes (strong verbatim signals)
+refsets = {}  # n -> set of shingles
+for n in (8, 12, 20, 30):
+    S = set()
+    for r in ref:
+        for fld in ('question','answer'):
+            sh,_ = shingles(norm(r[fld]), n)
+            S |= sh
+    refsets[n] = S
+print("ref ngram set sizes:", {k:len(v) for k,v in refsets.items()}, file=sys.stderr)
+
+inp = sys.argv[1]
+flag_count = defaultdict(int)
+flagged = []
+total = 0
+with open(inp) as f:
+    for li, line in enumerate(f):
+        line=line.strip()
+        if not line: continue
+        total += 1
+        try:
+            obj = json.loads(line)
+            if isinstance(obj, dict):
+                text = " \n ".join(str(v) for v in obj.values())
+            else:
+                text = str(obj)
+        except Exception:
+            text = line
+        tn = norm(text)
+        hits = 0
+        for n, S in refsets.items():
+            sh,_ = shingles(tn, n)
+            c = len(sh & S)
+            if c: flag_count[n]+=1
+            hits += c
+        if hits > 0:
+            flagged.append((li, hits, text[:200]))
+print(f"total docs: {total}")
+print("docs with >=1 matching ngram by n:", dict(flag_count))
+print("flagged docs (any n, n>=8):", len(flagged))
+for f in flagged[:20]:
+    print(" FLAG", f[0], "hits", f[1], repr(f[2][:120]))
+sys.exit(0 if not flagged else 1)

instances/end_of_run/d05/state/scripts/probe_next.py

@@ -0,0 +1,15 @@
+import sys, torch, json
+from transformers import AutoModelForCausalLM, AutoTokenizer
+path=sys.argv[1]
+tok=AutoTokenizer.from_pretrained(path,trust_remote_code=True)
+m=AutoModelForCausalLM.from_pretrained(path,dtype=torch.bfloat16,device_map="cuda",trust_remote_code=True); m.eval()
+rows=[json.loads(l) for i,l in enumerate(open("data/train_clean.jsonl")) if i<50]
+r=rows[0]
+for suffix in ["", r['completion'][:60]]:
+    prompt=f"<|im_start|>user\n{r['prompt']}<|im_end|>\n<|im_start|>assistant\n{suffix}"
+    ids=tok(prompt,return_tensors="pt").to("cuda")
+    with torch.no_grad(): logits=m(**ids).logits[0,-1].float()
+    top=torch.topk(logits,5)
+    print("suffix:",repr(suffix[:40]))
+    for v,i in zip(top.values.tolist(), top.indices.tolist()):
+        print(f"   {i:>7} {v:8.2f} {repr(tok.decode([i]))}")

instances/end_of_run/d05/state/scripts/train.py

@@ -0,0 +1,195 @@
+#!/usr/bin/env python3
+"""LoRA / full SFT for Qwen3-4B-Base on code instruction data (HumanEval-style format)."""
+import os, sys, json, math, random, argparse, time
+import torch
+from torch.utils.data import Dataset
+from transformers import AutoModelForCausalLM, AutoTokenizer, TrainingArguments, Trainer
+from peft import LoraConfig, get_peft_model
+
+TGT = ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"]
+MODEL = "Qwen/Qwen3-4B-Base"
+
+
+def parse_args():
+    p = argparse.ArgumentParser()
+    p.add_argument("--data", default="data/train_clean.jsonl")
+    p.add_argument("--out", default="out/lora1")
+    p.add_argument("--epochs", type=float, default=1.0)
+    p.add_argument("--max-samples", type=int, default=-1)
+    p.add_argument("--bs", type=int, default=8)
+    p.add_argument("--grad-accum", type=int, default=2)
+    p.add_argument("--lr", type=float, default=1e-4)
+    p.add_argument("--max-len", type=int, default=1536)
+    p.add_argument("--lora-r", type=int, default=64)
+    p.add_argument("--lora-alpha", type=int, default=128)
+    p.add_argument("--lora-dropout", type=float, default=0.05)
+    p.add_argument("--full", action="store_true")
+    p.add_argument("--max-steps", type=int, default=-1)
+    p.add_argument("--save-steps", type=int, default=100000)
+    p.add_argument("--seed", type=int, default=0)
+    p.add_argument("--pack", action="store_true")
+    p.add_argument("--no-gc", action="store_true")
+    p.add_argument("--term", default="<|im_end|>")
+    p.add_argument("--lm-head-lora", action="store_true")
+    return p.parse_args()
+
+
+def pack(items, max_len, seed=0):
+    random.seed(seed)
+    random.shuffle(items)
+    out = []
+    bi, bl = [], []
+    for ids, lab in items:
+        if bi and len(bi) + len(ids) > max_len:
+            out.append((bi, bl))
+            bi, bl = [], []
+        bi += ids
+        bl += lab
+    if bi:
+        out.append((bi, bl))
+    return out
+
+
+class SFTDataset(Dataset):
+    def __init__(self, items, max_len):
+        self.items = items
+        self.max_len = max_len
+
+    def __len__(self):
+        return len(self.items)
+
+    def __getitem__(self, i):
+        ids, labels = self.items[i]
+        if len(ids) > self.max_len:
+            ids = ids[: self.max_len]
+            labels = labels[: self.max_len]
+        return {"input_ids": ids, "labels": labels, "attention_mask": [1] * len(ids)}
+
+
+def pretokenize(rows, tok, max_len, term="<|im_end|>"):
+    items = []
+    for r in rows:
+        prompt = r["prompt"]
+        completion = r["completion"]
+        prefix = f"<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant\n"
+        suffix = f"{completion}{term}"
+        p_ids = tok(prefix, add_special_tokens=False)["input_ids"]
+        s_ids = tok(suffix, add_special_tokens=False)["input_ids"]
+        ids = p_ids + s_ids
+        labels = [-100] * len(p_ids) + s_ids
+        if len(ids) > max_len:
+            # keep the end of the completion (important tokens); drop from prompt
+            if len(s_ids) >= max_len:
+                ids = p_ids[:1] + s_ids[-(max_len - 1):]
+                labels = [-100] + s_ids[-(max_len - 1):]
+            else:
+                keep = max_len - len(s_ids)
+                ids = p_ids[:keep] + s_ids
+                labels = [-100] * keep + s_ids
+        items.append((ids, labels))
+    return items
+
+
+class Collator:
+    def __init__(self, pad_id):
+        self.pad_id = pad_id
+
+    def __call__(self, batch):
+        ml = max(len(b["input_ids"]) for b in batch)
+        input_ids, labels, attn = [], [], []
+        for b in batch:
+            n = ml - len(b["input_ids"])
+            input_ids.append(b["input_ids"] + [self.pad_id] * n)
+            labels.append(b["labels"] + [-100] * n)
+            attn.append(b["attention_mask"] + [0] * n)
+        return {
+            "input_ids": torch.tensor(input_ids, dtype=torch.long),
+            "labels": torch.tensor(labels, dtype=torch.long),
+            "attention_mask": torch.tensor(attn, dtype=torch.long),
+        }
+
+
+def main():
+    args = parse_args()
+    random.seed(args.seed)
+    tok = AutoTokenizer.from_pretrained(MODEL, trust_remote_code=True)
+    if tok.pad_token is None:
+        tok.pad_token = tok.eos_token
+
+    rows = []
+    with open(args.data) as f:
+        for line in f:
+            line = line.strip()
+            if line:
+                rows.append(json.loads(line))
+    random.shuffle(rows)
+    if args.max_samples > 0:
+        rows = rows[: args.max_samples]
+    print(f"training rows: {len(rows)}", flush=True)
+    t0t = time.time()
+    ds_items = pretokenize(rows, tok, args.max_len, term=args.term)
+    if args.pack:
+        ds_items = pack(ds_items, args.max_len, seed=args.seed)
+    print(f"pretokenized in {time.time()-t0t:.1f}s -> {len(ds_items)} sequences", flush=True)
+
+    model = AutoModelForCausalLM.from_pretrained(
+        MODEL, torch_dtype=torch.bfloat16, attn_implementation="flash_attention_2", trust_remote_code=True
+    )
+    model.config.use_cache = False
+    if not args.no_gc:
+        model.gradient_checkpointing_enable(gradient_checkpointing_kwargs={"use_reentrant": False})
+        model.enable_input_require_grads()
+
+    if not args.full:
+        lc = LoraConfig(
+            r=args.lora_r,
+            lora_alpha=args.lora_alpha,
+            lora_dropout=args.lora_dropout,
+            bias="none",
+            task_type="CAUSAL_LM",
+            target_modules=(TGT + (["embed_tokens", "lm_head"] if args.lm_head_lora else [])),
+        )
+        model = get_peft_model(model, lc)
+        model.print_trainable_parameters()
+
+    ds = SFTDataset(ds_items, args.max_len)
+    steps_per_epoch = max(1, len(rows) // (args.bs * args.grad_accum))
+    targs = TrainingArguments(
+        output_dir=args.out,
+        per_device_train_batch_size=args.bs,
+        gradient_accumulation_steps=args.grad_accum,
+        num_train_epochs=args.epochs,
+        max_steps=args.max_steps,
+        learning_rate=args.lr,
+        lr_scheduler_type="cosine",
+        warmup_ratio=0.03,
+        weight_decay=0.0,
+        bf16=True,
+        logging_steps=20,
+        save_steps=args.save_steps,
+        save_total_limit=2,
+        report_to=[],
+        dataloader_num_workers=4,
+        gradient_checkpointing=False,
+        optim="adamw_torch",
+        seed=args.seed,
+    )
+    trainer = Trainer(
+        model=model,
+        args=targs,
+        train_dataset=ds,
+        data_collator=Collator(tok.pad_token_id),
+    )
+    t0 = time.time()
+    trainer.train()
+    print("train time (s):", time.time() - t0, flush=True)
+
+    # save
+    os.makedirs(args.out, exist_ok=True)
+    model.save_pretrained(args.out)
+    tok.save_pretrained(args.out)
+    print("saved to", args.out, flush=True)
+
+
+if __name__ == "__main__":
+    main()

instances/end_of_run/d05/state/scripts/verify_clean.py

@@ -0,0 +1,24 @@
+import json, re
+def norm(s):
+    s=s.lower(); s=re.sub(r'[^a-z0-9_]+',' ',s); return re.sub(r'\s+',' ',s).strip()
+def sh(text,n):
+    t=text.split(); return set(tuple(t[i:i+n]) for i in range(max(0,len(t)-n+1)))
+ref=json.load(open('/mnt/ben/test_data.json'))
+R=[]
+for i,r in enumerate(ref):
+    for fld in ('question','answer'):
+        R.append((i,fld,norm(r[fld]),sh(norm(r[fld]),12),sh(norm(r[fld]),20)))
+mx=0; n12=0; n20=0; ex=None
+tot=0
+with open('data/train_clean.jsonl') as f:
+    for line in f:
+        obj=json.loads(line); tot+=1
+        tn=norm(obj['prompt']+" \n "+obj['completion'])
+        ts12=sh(tn,12); ts20=sh(tn,20)
+        for i,fld,rtxt,s12,s20 in R:
+            c12=len(ts12&s12); c20=len(ts20&s20)
+            if c20: n20+=1
+            if c12>=2: n12+=1
+            if c12+5*c20>mx: mx=c12+5*c20; ex=(i,fld,c12,c20,obj['prompt'][:80])
+print("total",tot,"docs with >=2 shared 12-grams w/ single ref:",n12,"docs with any 20-gram match:",n20)
+print("max score:",mx, ex)

instances/end_of_run/d05/state/templates/gemma3.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

instances/end_of_run/d05/state/templates/gemma3_tool_calling.jinja

@@ -0,0 +1,143 @@
+{{ bos_token }}
+{# ------------------------------------------------------------
+   1) TOOLS / SYSTEM BLOCK
+   If tools are provided, we emit a system turn that:
+   - includes the original system message (if any)
+   - appends tool signatures inside <tools>...</tools>
+   - explains how to return tool calls
+   ------------------------------------------------------------ #}
+{%- set has_system = messages and messages[0]['role'] == 'system' -%}
+{%- if tools %}
+<start_of_turn>system
+{%- if has_system -%}
+{{ messages[0]['content'] }}
+
+{%- endif -%}
+# Tools
+
+You may call one or more functions to assist with the user query.
+
+You are provided with function signatures within <tools></tools> XML tags:
+<tools>
+{%- for tool in tools %}
+{{ tool | tojson }}
+{%- if not loop.last %}
+
+{%- endif %}
+{%- endfor %}
+</tools>
+
+For each function call, return a JSON object with the function name and arguments
+inside <tool_call></tool_call> tags, like:
+
+<tool_call>
+{"name": "<function-name>", "arguments": { ... }}
+</tool_call>
+<end_of_turn>
+{%- endif %}
+
+{# ------------------------------------------------------------
+   2) DETERMINE WHICH MESSAGES TO LOOP OVER
+   - If we already consumed the system message in the tools block, skip it.
+   - Otherwise, keep Gemma's original "first_user_prefix" behavior.
+   ------------------------------------------------------------ #}
+{%- if tools %}
+    {# tools block already printed system, so skip the first system (if present) #}
+    {%- if has_system -%}
+        {%- set first_user_prefix = "" -%}
+        {%- set loop_messages = messages[1:] -%}
+    {%- else -%}
+        {%- set first_user_prefix = "" -%}
+        {%- set loop_messages = messages -%}
+    {%- endif -%}
+{%- else %}
+    {# original Gemma behavior #}
+    {%- if has_system -%}
+        {%- if messages[0]['content'] is string -%}
+            {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+        {%- else -%}
+            {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+        {%- endif -%}
+        {%- set loop_messages = messages[1:] -%}
+    {%- else -%}
+        {%- set first_user_prefix = "" -%}
+        {%- set loop_messages = messages -%}
+    {%- endif -%}
+{%- endif %}
+
+{# ------------------------------------------------------------
+   3) MESSAGE LOOP
+   We keep Gemma's <start_of_turn> / <end_of_turn> format,
+   but relax strict alternation to allow 'tool' turns.
+   - user -> <start_of_turn>user ... <end_of_turn>
+   - assistant -> <start_of_turn>model ... <end_of_turn>
+   - tool -> wrapped as a user turn with <tool_response>...</tool_response>
+     (this is the pattern Qwen uses, and it lets the model read tool outputs)
+   ------------------------------------------------------------ #}
+{%- for message in loop_messages -%}
+    {# determine role text #}
+    {%- if message['role'] == 'assistant' -%}
+        {%- set role = 'model' -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+
+    {# TOOL RESPONSE MESSAGES
+       We render tool messages as a user turn that contains <tool_response>...</tool_response>
+       so the model can consume the tool outputs on the next assistant turn.
+    #}
+    {%- if message['role'] == 'tool' -%}
+<start_of_turn>user
+<tool_response>
+{%- if message['content'] is string -%}
+{{ message['content'] | trim }}
+{%- else -%}
+{{ message['content'] | tojson }}
+{%- endif -%}
+</tool_response>
+<end_of_turn>
+    {%- else -%}
+        {# NORMAL USER / ASSISTANT / SYSTEM (non-initial) MESSAGE #}
+<start_of_turn>{{ role }}
+{%- if loop.first and first_user_prefix -%}
+{{ first_user_prefix }}
+{%- endif -%}
+        {%- if message['content'] is string -%}
+{{ message['content'] | trim }}
+        {%- elif message['content'] is iterable -%}
+            {%- for item in message['content'] -%}
+                {%- if item['type'] == 'image' -%}
+<start_of_image>
+                {%- elif item['type'] == 'text' -%}
+{{ item['text'] | trim }}
+                {%- endif -%}
+            {%- endfor -%}
+        {%- else -%}
+{{ raise_exception("Invalid content type") }}
+        {%- endif -%}
+
+        {# If this assistant message includes tool calls, append them here #}
+        {%- if message['role'] == 'assistant' and message.get('tool_calls') -%}
+            {%- for tool_call in message['tool_calls'] -%}
+                {# unify format: tool_call.function may exist #}
+                {%- set tc = tool_call.function if tool_call.get('function') else tool_call -%}
+
+<tool_call>
+{"name": "{{ tc.name }}", "arguments": {{ tc.arguments if tc.arguments is string else (tc.arguments | tojson) }}}
+</tool_call>
+            {%- endfor -%}
+        {%- endif -%}
+<end_of_turn>
+    {%- endif -%}
+{%- endfor -%}
+
+{# ------------------------------------------------------------
+   4) GENERATION PROMPT
+   ------------------------------------------------------------ #}
+{%- if add_generation_prompt -%}
+<start_of_turn>model
+{%- endif -%}

instances/end_of_run/d05/state/templates/qwen3.jinja

@@ -0,0 +1,89 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n" }}{{ '{"name": <function-name>, "arguments": <args-json-object>}' }}{{- "\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is string %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in content %}
+                {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+                {%- set content = content.split('</think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

instances/end_of_run/d05/state/templates/smollm.jinja

@@ -0,0 +1,94 @@
+{# ───── defaults ───── #}
+{%- if enable_thinking is not defined -%}
+{%- set enable_thinking = true -%}
+{%- endif -%}
+
+{# ───── reasoning mode ───── #}
+{%- if enable_thinking -%}
+  {%- set reasoning_mode = "/think" -%}
+{%- else -%}
+  {%- set reasoning_mode = "/no_think" -%}
+{%- endif -%}
+
+{# ───── header (system message) ───── #}
+{{- "<|im_start|>system\n" -}}
+
+{%- if messages[0].role == "system" -%}
+  {%- set system_message = messages[0].content -%}
+  {%- if "/no_think" in system_message -%}
+    {%- set reasoning_mode = "/no_think" -%}
+  {%- elif "/think" in system_message -%}
+    {%- set reasoning_mode = "/think" -%}
+  {%- endif -%}
+  {%- set custom_instructions = system_message.replace("/no_think", "").replace("/think", "").rstrip() -%}
+{%- endif -%}
+
+{%- if "/system_override" in system_message -%}
+  {{- custom_instructions.replace("/system_override", "").rstrip() -}}
+  {{- "<|im_end|>\n" -}}
+{%- else -%}
+  {{- "## Metadata\n\n" -}}
+  {{- "Knowledge Cutoff Date: June 2025\n" -}}
+  {%- set today = strftime_now("%d %B %Y") -%}
+  {{- "Today Date: " ~ today ~ "\n" -}}
+  {{- "Reasoning Mode: " + reasoning_mode + "\n\n" -}}
+  
+  {{- "## Custom Instructions\n\n" -}}
+  {%- if custom_instructions -%}
+    {{- custom_instructions + "\n\n" -}}
+  {%- elif reasoning_mode == "/think" -%}
+    {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracking, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format: <think> Thought section </think> Solution section. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion.\n\n" -}}
+  {%- else -%}
+    {{- "You are a helpful AI assistant named SmolLM, trained by Hugging Face.\n\n" -}}
+  {%- endif -%}
+
+  {%- if xml_tools or python_tools or tools -%}
+    {{- "### Tools\n\n" -}}
+    {%- if xml_tools or tools -%}
+      {%- if tools -%}
+        {%- set xml_tools = tools -%}
+      {%- endif -%}
+      {%- set ns = namespace(xml_tool_string="You may call one or more functions to assist with the user query.\nYou are provided with function signatures within <tools></tools> XML tags:\n\n<tools>\n") -%}
+      {%- for tool in xml_tools[:] -%} {# The slicing makes sure that xml_tools is a list #}
+        {%- set ns.xml_tool_string = ns.xml_tool_string ~ (tool | string) ~ "\n" -%}
+      {%- endfor -%}
+      {%- set xml_tool_string = ns.xml_tool_string + "</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call>" -%}
+      {{- xml_tool_string -}}
+    {%- endif -%}
+    {%- if python_tools -%}
+      {%- set ns = namespace(python_tool_string="When you send a message containing Python code between '<code>' and '</code>' tags, it will be executed in a stateful Jupyter notebook environment, and you will then be given the output to continued reasoning in an agentic loop.\n\nYou can use the following tools in your python code like regular functions:\n<tools>\n") -%}
+      {%- for tool in python_tools[:] -%} {# The slicing makes sure that python_tools is a list #}
+        {%- set ns.python_tool_string = ns.python_tool_string ~ (tool | string) ~ "\n" -%}
+      {%- endfor -%}
+      {%- set python_tool_string = ns.python_tool_string + "</tools>\n\nThe state persists between code executions: so variables that you define in one step are still available thereafter." -%}
+      {{- python_tool_string -}}
+    {%- endif -%}
+    {{- "\n\n" -}}
+    {{- "<|im_end|>\n" -}}
+  {%- endif -%}
+{%- endif -%}
+{# ───── main loop ───── #}
+{%- for message in messages -%}
+    {%- set content = message.content if message.content is string else "" -%}
+    {%- if message.role == "user" -%}
+        {{ "<|im_start|>" + message.role + "\n"  + content + "<|im_end|>\n" }}
+    {%- elif message.role == "assistant" -%}
+        {% generation %}
+        {%- if reasoning_mode == "/think" -%}
+            {{ "<|im_start|>assistant\n" + content.lstrip("\n") + "<|im_end|>\n" }}
+        {%- else -%}
+            {{ "<|im_start|>assistant\n" + "<think>\n\n</think>\n" + content.lstrip("\n") + "<|im_end|>\n" }}
+        {%- endif -%}
+        {% endgeneration %}
+    {%- elif message.role == "tool" -%}
+    {{ "<|im_start|>" + "user\n"  + content + "<|im_end|>\n" }}
+    {%- endif -%}
+{%- endfor -%}
+{# ───── generation prompt ───── #}
+{%- if add_generation_prompt -%}
+    {%- if reasoning_mode == "/think" -%}
+        {{ "<|im_start|>assistant\n" }}
+    {%- else -%}
+        {{ "<|im_start|>assistant\n" + "<think>\n\n</think>\n"  }}
+    {%- endif -%}
+{%- endif -%}

instances/end_of_run/d05/state/timer.sh

@@ -0,0 +1,17 @@
+#!/bin/bash
+
+NUM_HOURS=5
+CREATION_DATE=1793019139
+
+DEADLINE=$((CREATION_DATE + NUM_HOURS * 3600))
+NOW=$(date +%s)
+REMAINING=$((DEADLINE - NOW))
+
+if [ $REMAINING -le 0 ]; then
+    echo "Timer expired!"
+else
+    echo "Remaining time (hours:minutes)":
+    HOURS=$((REMAINING / 3600))
+    MINUTES=$(((REMAINING % 3600) / 60))
+    printf "%d:%02d\n" $HOURS $MINUTES
+fi