初始化项目，由ModelHub XC社区提供模型

Model: jy1095/qwen3-0.6b-neucodec-multipack-test
Source: Original Platform

.gitattributes

@@ -0,0 +1,37 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
+train_loss_curve.png filter=lfs diff=lfs merge=lfs -text

chat_template.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 %}

config.json

@@ -0,0 +1,63 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "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"
+  ],
+  "max_position_embeddings": 40960,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.12.1",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 217207
+}

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "bos_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "temperature": 0.6,
+  "top_k": 20,
+  "top_p": 0.95,
+  "transformers_version": "5.12.1"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6e46152b1eea4cdfe06d8e5ae278af029dd5e0ebfcd6f3f9167694baf322d42
+size 1325810200

qwen_train_subset_multipack.py

@@ -0,0 +1,275 @@
+import csv
+import io
+import zipfile
+from pathlib import Path
+
+import pandas as pd
+import torch
+from torch.optim import AdamW
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+
+DATASET_DIR = Path("/workspace/fleurs-r-neucodec")
+MODEL_NAME = "Qwen/Qwen3-0.6B"
+
+NUM_SPEECH_TOKENS = 65536
+MAX_SPEECH_TOKENS = 500
+MAX_LENGTH = 1536
+
+TRAIN_SPLIT = "train"
+VAL_SPLIT = "dev"
+
+MAX_TRAIN_EXAMPLES = 500
+MAX_VAL_EXAMPLES = 50
+
+LR = 1e-5
+EPOCHS = 1
+EVAL_EVERY = 25
+SAVE_DIR = Path("/workspace/qwen_speech_multipack_2_ckpt")
+LOG_CSV = Path("/workspace/train_log_multipack_2.csv")
+
+
+def list_token_zips(split):
+    zips = sorted((DATASET_DIR / "neucodec").glob(f"en_us-{split}*.zip"))
+    if not zips:
+        raise FileNotFoundError(f"No token zips found for split={split}")
+    return zips
+
+
+def build_zip_index(zip_paths):
+    index = {}
+    open_zips = []
+
+    for path in zip_paths:
+        zf = zipfile.ZipFile(path)
+        open_zips.append(zf)
+
+        for name in zf.namelist():
+            if name.endswith(".pt"):
+                stem = Path(name).stem
+                index[stem] = (zf, name)
+
+    return index, open_zips
+
+
+def load_codes(zip_index, neucodec_path):
+    stem = Path(str(neucodec_path).replace("\\", "/")).stem
+
+    if stem not in zip_index:
+        raise FileNotFoundError(f"No token file found for {neucodec_path}")
+
+    zf, entry = zip_index[stem]
+    obj = torch.load(io.BytesIO(zf.read(entry)), map_location="cpu")
+    return obj["codes"].flatten().to(torch.long).tolist()
+
+
+def build_single_example(tokenizer, codes, transcript):
+    codes = codes[:MAX_SPEECH_TOKENS]
+
+    speech_text = " ".join(f"<speech_{code}>" for code in codes)
+    prompt = f"<speech_start> {speech_text} <speech_end>\n"
+    target = str(transcript) + tokenizer.eos_token
+
+    prompt_ids = tokenizer(prompt, add_special_tokens=False)["input_ids"]
+    target_ids = tokenizer(target, add_special_tokens=False)["input_ids"]
+
+    input_ids = prompt_ids + target_ids
+    labels = [-100] * len(prompt_ids) + target_ids
+
+    input_ids = input_ids[:MAX_LENGTH]
+    labels = labels[:MAX_LENGTH]
+
+    return input_ids, labels
+
+
+def pack_examples(single_examples):
+    packed = []
+    cur_input_ids = []
+    cur_labels = []
+    cur_segment_ids = []
+    segment_id = 0
+
+    for input_ids, labels in single_examples:
+        if not input_ids:
+            continue
+
+        if cur_input_ids and len(cur_input_ids) + len(input_ids) > MAX_LENGTH:
+            packed.append(
+                {
+                    "input_ids": torch.tensor(cur_input_ids, dtype=torch.long),
+                    "labels": torch.tensor(cur_labels, dtype=torch.long),
+                    "segment_ids": torch.tensor(cur_segment_ids, dtype=torch.long),
+                }
+            )
+            cur_input_ids = []
+            cur_labels = []
+            cur_segment_ids = []
+            segment_id = 0
+
+        if len(input_ids) > MAX_LENGTH:
+            input_ids = input_ids[:MAX_LENGTH]
+            labels = labels[:MAX_LENGTH]
+
+        cur_input_ids.extend(input_ids)
+        cur_labels.extend(labels)
+        cur_segment_ids.extend([segment_id] * len(input_ids))
+        segment_id += 1
+
+    if cur_input_ids:
+        packed.append(
+            {
+                "input_ids": torch.tensor(cur_input_ids, dtype=torch.long),
+                "labels": torch.tensor(cur_labels, dtype=torch.long),
+                "segment_ids": torch.tensor(cur_segment_ids, dtype=torch.long),
+            }
+        )
+
+    return packed
+
+
+def load_examples(tokenizer, split, max_examples):
+    parquet = DATASET_DIR / "data" / f"en_us-{split}.parquet"
+    df = pd.read_parquet(parquet).head(max_examples)
+
+    zip_paths = list_token_zips(split)
+    zip_index, open_zips = build_zip_index(zip_paths)
+
+    single_examples = []
+    for _, row in df.iterrows():
+        codes = load_codes(zip_index, row["neucodec_path"])
+        single_examples.append(build_single_example(tokenizer, codes, row["sentence"]))
+
+    packed_examples = pack_examples(single_examples)
+    return packed_examples, open_zips
+
+
+def make_block_causal_mask(segment_ids, dtype):
+    # segment_ids: [L]. Tokens can attend only to earlier tokens in the same packed example.
+    segment_ids = segment_ids.cuda()
+    length = segment_ids.numel()
+    same_segment = segment_ids[:, None] == segment_ids[None, :]
+    causal = torch.arange(length, device="cuda")[:, None] >= torch.arange(length, device="cuda")[None, :]
+    allowed = same_segment & causal
+
+    mask = torch.zeros((1, 1, length, length), device="cuda", dtype=dtype)
+    mask = mask.masked_fill(~allowed[None, None, :, :], torch.finfo(dtype).min)
+    return mask
+
+
+def make_position_ids(segment_ids):
+    # Reset positions at each packed-example boundary.
+    position_ids = torch.zeros_like(segment_ids)
+    for segment in torch.unique(segment_ids):
+        idx = torch.nonzero(segment_ids == segment, as_tuple=False).flatten()
+        position_ids[idx] = torch.arange(idx.numel(), dtype=torch.long)
+    return position_ids.unsqueeze(0).cuda()
+
+
+@torch.inference_mode()
+def evaluate(model, examples):
+    model.eval()
+    losses = []
+
+    for ex in examples:
+        input_ids = ex["input_ids"].unsqueeze(0).cuda()
+        labels = ex["labels"].unsqueeze(0).cuda()
+        attention_mask = make_block_causal_mask(ex["segment_ids"], model.dtype)
+        position_ids = make_position_ids(ex["segment_ids"])
+
+        out = model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            labels=labels,
+        )
+        losses.append(float(out.loss))
+
+    model.train()
+    return sum(losses) / len(losses)
+
+
+def main():
+    tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, trust_remote_code=True)
+    tokenizer.pad_token = tokenizer.eos_token
+
+    speech_tokens = [f"<speech_{i}>" for i in range(NUM_SPEECH_TOKENS)]
+    tokenizer.add_tokens(["<speech_start>", "<speech_end>"] + speech_tokens)
+
+    print("Loading and multipacking train examples...")
+    train_examples, train_zips = load_examples(tokenizer, TRAIN_SPLIT, MAX_TRAIN_EXAMPLES)
+
+    print("Loading and multipacking validation examples...")
+    val_examples, val_zips = load_examples(tokenizer, VAL_SPLIT, MAX_VAL_EXAMPLES)
+
+    print(f"packed train batches: {len(train_examples)}")
+    print(f"packed val batches: {len(val_examples)}")
+    print(f"vocab size: {len(tokenizer)}")
+
+    model = AutoModelForCausalLM.from_pretrained(
+        MODEL_NAME,
+        torch_dtype=torch.bfloat16,
+        trust_remote_code=True,
+    )
+    model.resize_token_embeddings(len(tokenizer))
+    model.cuda()
+    model.train()
+
+    optimizer = AdamW(model.parameters(), lr=LR)
+
+    with LOG_CSV.open("w", newline="") as f:
+        writer = csv.DictWriter(f, fieldnames=["step", "train_loss", "val_loss"])
+        writer.writeheader()
+
+    step = 0
+    for epoch in range(EPOCHS):
+        for ex in train_examples:
+            step += 1
+
+            input_ids = ex["input_ids"].unsqueeze(0).cuda()
+            labels = ex["labels"].unsqueeze(0).cuda()
+            attention_mask = make_block_causal_mask(ex["segment_ids"], model.dtype)
+            position_ids = make_position_ids(ex["segment_ids"])
+
+            out = model(
+                input_ids=input_ids,
+                attention_mask=attention_mask,
+                position_ids=position_ids,
+                labels=labels,
+            )
+
+            loss = out.loss
+            loss.backward()
+            optimizer.step()
+            optimizer.zero_grad(set_to_none=True)
+
+            train_loss = float(loss.detach())
+            val_loss = ""
+
+            if step % EVAL_EVERY == 0:
+                val_loss = evaluate(model, val_examples)
+                print(f"step {step:04d} train_loss {train_loss:.4f} val_loss {val_loss:.4f}")
+            else:
+                print(f"step {step:04d} train_loss {train_loss:.4f}")
+
+            with LOG_CSV.open("a", newline="") as f:
+                writer = csv.DictWriter(f, fieldnames=["step", "train_loss", "val_loss"])
+                writer.writerow(
+                    {
+                        "step": step,
+                        "train_loss": train_loss,
+                        "val_loss": val_loss,
+                    }
+                )
+
+    SAVE_DIR.mkdir(parents=True, exist_ok=True)
+    model.save_pretrained(SAVE_DIR)
+    tokenizer.save_pretrained(SAVE_DIR)
+    print(f"saved checkpoint: {SAVE_DIR}")
+    print(f"saved log: {LOG_CSV}")
+
+    for zf in train_zips + val_zips:
+        zf.close()
+
+
+if __name__ == "__main__":
+    main()

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6683f1da5a789077b380d431d64b639063feb7b51e394e6eab15fb2e2ae3072
+size 23929296

tokenizer_config.json

@@ -0,0 +1,30 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_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|>"
+  ],
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 131072,
+  "pad_token": "<|im_end|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

train_log_multipack_2.csv

@@ -0,0 +1,459 @@
+step,train_loss,val_loss
+1,4.83737850189209,
+2,4.579104423522949,
+3,3.8569114208221436,
+4,4.565186977386475,
+5,3.4103474617004395,
+6,4.785758018493652,
+7,4.017621994018555,
+8,4.826681137084961,
+9,3.9895143508911133,
+10,4.845824241638184,
+11,4.819087028503418,
+12,4.493717670440674,
+13,2.8794360160827637,
+14,4.196484565734863,
+15,3.2648439407348633,
+16,4.475448131561279,
+17,3.101593017578125,
+18,3.4317128658294678,
+19,4.627690315246582,
+20,3.764618396759033,
+21,4.100680351257324,
+22,4.149367332458496,
+23,1.7279186248779297,
+24,3.528687000274658,
+25,4.040360450744629,3.6418914389103016
+26,2.5350730419158936,
+27,4.534997463226318,
+28,4.401715278625488,
+29,3.060065507888794,
+30,3.409668207168579,
+31,3.075878143310547,
+32,3.9714224338531494,
+33,2.8314208984375,
+34,3.813950300216675,
+35,3.7870593070983887,
+36,3.8533389568328857,
+37,3.509331703186035,
+38,4.414717674255371,
+39,3.541666030883789,
+40,3.0279834270477295,
+41,2.312675714492798,
+42,3.2705774307250977,
+43,3.316633939743042,
+44,4.442257404327393,
+45,4.468428134918213,
+46,3.3701953887939453,
+47,4.127744197845459,
+48,3.1203060150146484,
+49,3.164045810699463,
+50,3.515629529953003,3.555581473289652
+51,4.224531173706055,
+52,3.420295476913452,
+53,3.550535202026367,
+54,3.0400187969207764,
+55,4.082877159118652,
+56,2.9616825580596924,
+57,3.734877347946167,
+58,5.549570083618164,
+59,3.850841999053955,
+60,4.226750373840332,
+61,4.564910411834717,
+62,3.049321174621582,
+63,3.783742666244507,
+64,3.2545578479766846,
+65,3.968663215637207,
+66,4.826290130615234,
+67,3.6602656841278076,
+68,4.045124530792236,
+69,3.5444107055664062,
+70,3.9055728912353516,
+71,2.661231517791748,
+72,3.7383553981781006,
+73,3.1351959705352783,
+74,4.062368392944336,
+75,3.156069755554199,3.514297257078455
+76,3.0109686851501465,
+77,4.426510810852051,
+78,3.423661470413208,
+79,3.532076835632324,
+80,3.2007856369018555,
+81,2.7224533557891846,
+82,3.541182518005371,
+83,4.03301477432251,
+84,3.0639500617980957,
+85,2.8058218955993652,
+86,4.198553562164307,
+87,4.709975242614746,
+88,3.4877724647521973,
+89,3.526134967803955,
+90,3.586360454559326,
+91,2.8046786785125732,
+92,3.2414307594299316,
+93,3.693235158920288,
+94,3.6420347690582275,
+95,4.170336723327637,
+96,2.8600430488586426,
+97,3.4960365295410156,
+98,2.7975122928619385,
+99,4.142976760864258,
+100,3.742992401123047,3.487763815737785
+101,3.799506187438965,
+102,2.9835662841796875,
+103,3.813058614730835,
+104,3.8280656337738037,
+105,3.2202537059783936,
+106,3.07291316986084,
+107,3.071430206298828,
+108,3.132784128189087,
+109,4.195946216583252,
+110,2.929187059402466,
+111,3.574751377105713,
+112,3.8678948879241943,
+113,3.149278163909912,
+114,3.6508023738861084,
+115,4.515976428985596,
+116,4.646908760070801,
+117,4.965893268585205,
+118,2.782670497894287,
+119,4.051075458526611,
+120,3.78477144241333,
+121,3.0531561374664307,
+122,3.8409078121185303,
+123,3.7307589054107666,
+124,4.737176895141602,
+125,2.0638153553009033,3.4708150904229345
+126,2.3227977752685547,
+127,3.319139003753662,
+128,4.758574485778809,
+129,4.331243515014648,
+130,3.439526081085205,
+131,3.6598761081695557,
+132,4.913437366485596,
+133,4.259307384490967,
+134,2.245208263397217,
+135,3.726602792739868,
+136,4.2422943115234375,
+137,3.2248215675354004,
+138,3.0694921016693115,
+139,4.343524932861328,
+140,2.5264651775360107,
+141,4.922786712646484,
+142,3.552476406097412,
+143,3.1056058406829834,
+144,4.8071675300598145,
+145,1.5579023361206055,
+146,3.97298002243042,
+147,3.3424508571624756,
+148,3.5528564453125,
+149,3.0181264877319336,
+150,3.5836517810821533,3.464779980639194
+151,4.640353679656982,
+152,3.947347402572632,
+153,3.9362823963165283,
+154,4.109447956085205,
+155,3.775275468826294,
+156,2.2141740322113037,
+157,3.9296300411224365,
+158,3.847964286804199,
+159,3.5097544193267822,
+160,3.119296073913574,
+161,3.451831102371216,
+162,3.1743721961975098,
+163,3.4725611209869385,
+164,4.318027973175049,
+165,3.184769630432129,
+166,3.5490171909332275,
+167,3.897948980331421,
+168,3.4800829887390137,
+169,3.035662889480591,
+170,3.121901512145996,
+171,2.8668429851531982,
+172,2.8848462104797363,
+173,3.2499208450317383,
+174,2.87648868560791,
+175,3.1791110038757324,3.45171933985771
+176,3.20145845413208,
+177,3.6656036376953125,
+178,4.254746913909912,
+179,3.8568413257598877,
+180,3.3873519897460938,
+181,2.5699331760406494,
+182,4.006359577178955,
+183,4.802577018737793,
+184,3.8868982791900635,
+185,3.3518261909484863,
+186,2.449648857116699,
+187,4.026482105255127,
+188,3.218484878540039,
+189,2.801923990249634,
+190,2.4700212478637695,
+191,4.033946990966797,
+192,3.5595896244049072,
+193,2.2776455879211426,
+194,5.420859336853027,
+195,2.804280996322632,
+196,2.85151743888855,
+197,2.7877354621887207,
+198,3.6410443782806396,
+199,2.140655994415283,
+200,4.907998561859131,3.4415235874500683
+201,3.9623546600341797,
+202,3.770249605178833,
+203,3.7238171100616455,
+204,3.3219447135925293,
+205,2.794365882873535,
+206,3.19931960105896,
+207,4.658032417297363,
+208,4.272351264953613,
+209,3.0018627643585205,
+210,4.852128982543945,
+211,4.100520133972168,
+212,3.228710651397705,
+213,2.8996849060058594,
+214,4.030970573425293,
+215,3.3683249950408936,
+216,3.934347152709961,
+217,3.7863426208496094,
+218,3.919623613357544,
+219,4.968966007232666,
+220,4.764462947845459,
+221,2.889967441558838,
+222,4.756126880645752,
+223,3.971885919570923,
+224,4.080149173736572,
+225,4.033597469329834,3.4432580166674676
+226,4.219792366027832,
+227,2.8466875553131104,
+228,3.5342724323272705,
+229,3.143789768218994,
+230,2.317599296569824,
+231,3.4089839458465576,
+232,3.8498101234436035,
+233,2.375635862350464,
+234,4.283705234527588,
+235,3.4035141468048096,
+236,4.934589862823486,
+237,3.1188302040100098,
+238,2.789722442626953,
+239,1.6042765378952026,
+240,3.65496563911438,
+241,4.631184101104736,
+242,3.5723822116851807,
+243,4.454005718231201,
+244,3.399834156036377,
+245,3.7455062866210938,
+246,4.557286262512207,
+247,3.1282284259796143,
+248,3.116020917892456,
+249,3.6729848384857178,
+250,3.3174736499786377,3.444297293399243
+251,2.4893667697906494,
+252,3.791905403137207,
+253,4.210204601287842,
+254,3.109525680541992,
+255,2.527846336364746,
+256,4.1202006340026855,
+257,4.2178826332092285,
+258,3.3063230514526367,
+259,2.333925485610962,
+260,5.110389232635498,
+261,2.777125597000122,
+262,3.2000536918640137,
+263,3.6621885299682617,
+264,4.39784574508667,
+265,3.153855562210083,
+266,5.073533058166504,
+267,2.8840157985687256,
+268,3.6498281955718994,
+269,2.7056210041046143,
+270,4.258342742919922,
+271,4.288335800170898,
+272,3.777733564376831,
+273,2.9544053077697754,
+274,4.371647357940674,
+275,3.1846938133239746,3.4193343050936433
+276,3.6150033473968506,
+277,3.0862834453582764,
+278,2.6581320762634277,
+279,3.1265101432800293,
+280,2.7327654361724854,
+281,4.979248046875,
+282,3.249157190322876,
+283,3.3512628078460693,
+284,4.189081192016602,
+285,4.366570472717285,
+286,3.923560857772827,
+287,3.81119441986084,
+288,4.06733512878418,
+289,3.2456679344177246,
+290,3.121525764465332,
+291,4.4502763748168945,
+292,2.859525442123413,
+293,3.103595018386841,
+294,3.5803730487823486,
+295,3.3084216117858887,
+296,5.044394493103027,
+297,4.349173545837402,
+298,2.487546443939209,
+299,3.441528081893921,
+300,4.484344005584717,3.422133039920888
+301,3.191145896911621,
+302,3.6174354553222656,
+303,3.4205496311187744,
+304,3.1575920581817627,
+305,2.7539448738098145,
+306,2.5054097175598145,
+307,4.509216785430908,
+308,4.329964637756348,
+309,3.087510347366333,
+310,3.219388723373413,
+311,2.9950687885284424,
+312,3.3501241207122803,
+313,3.3984124660491943,
+314,3.0490026473999023,
+315,3.440187931060791,
+316,5.148359298706055,
+317,3.3574347496032715,
+318,2.603912591934204,
+319,3.200441360473633,
+320,2.2681984901428223,
+321,1.3850171566009521,
+322,4.180147647857666,
+323,2.865475654602051,
+324,3.8928990364074707,
+325,2.913665294647217,3.4212434951295245
+326,2.9794881343841553,
+327,3.4925873279571533,
+328,2.337425708770752,
+329,3.538888454437256,
+330,3.5051937103271484,
+331,4.471581935882568,
+332,2.1943535804748535,
+333,2.4782488346099854,
+334,4.359781742095947,
+335,3.9852702617645264,
+336,3.104440450668335,
+337,2.7623770236968994,
+338,2.6553149223327637,
+339,2.8583669662475586,
+340,3.085019111633301,
+341,2.5710747241973877,
+342,4.485040664672852,
+343,2.7791786193847656,
+344,3.3884494304656982,
+345,2.7618203163146973,
+346,3.5326671600341797,
+347,3.1701102256774902,
+348,2.5622284412384033,
+349,3.6489336490631104,
+350,4.041916847229004,3.417217533639137
+351,3.303135633468628,
+352,3.7105326652526855,
+353,2.986264228820801,
+354,3.894716501235962,
+355,3.408475160598755,
+356,2.84619140625,
+357,3.485544443130493,
+358,3.0264556407928467,
+359,3.6007182598114014,
+360,3.5983495712280273,
+361,3.245755195617676,
+362,2.8272345066070557,
+363,2.6484899520874023,
+364,3.5493357181549072,
+365,3.6680004596710205,
+366,4.227343559265137,
+367,3.180736780166626,
+368,4.276458263397217,
+369,3.379707098007202,
+370,3.913658618927002,
+371,3.6986403465270996,
+372,4.247578144073486,
+373,2.886542320251465,
+374,4.0589470863342285,
+375,4.18842887878418,3.4187491853186427
+376,4.287686347961426,
+377,3.424771785736084,
+378,4.727417945861816,
+379,2.1747705936431885,
+380,3.835639476776123,
+381,1.264375925064087,
+382,3.4173645973205566,
+383,3.6598403453826904,
+384,3.8715529441833496,
+385,3.3528504371643066,
+386,5.078052997589111,
+387,2.4622654914855957,
+388,3.9295761585235596,
+389,4.596060276031494,
+390,4.146194934844971,
+391,3.3156752586364746,
+392,2.829684019088745,
+393,3.097052812576294,
+394,2.0725510120391846,
+395,2.5597586631774902,
+396,3.6028549671173096,
+397,3.8087193965911865,
+398,3.8290321826934814,
+399,3.3390116691589355,
+400,2.952338695526123,3.4204193125379847
+401,4.4381818771362305,
+402,3.758164644241333,
+403,3.729465961456299,
+404,3.8695130348205566,
+405,4.027613162994385,
+406,3.333869695663452,
+407,3.0652832984924316,
+408,3.888929605484009,
+409,3.106435775756836,
+410,3.2263383865356445,
+411,3.8024957180023193,
+412,2.508396625518799,
+413,3.454288959503174,
+414,4.009916305541992,
+415,4.484684944152832,
+416,3.197667121887207,
+417,3.8992998600006104,
+418,2.119562864303589,
+419,3.722830295562744,
+420,3.4790313243865967,
+421,3.4761900901794434,
+422,2.4494388103485107,
+423,3.956528902053833,
+424,3.0699243545532227,
+425,3.4583213329315186,3.412918080674841
+426,3.8467800617218018,
+427,3.8987865447998047,
+428,5.462136268615723,
+429,4.194992542266846,
+430,3.279252529144287,
+431,3.5070934295654297,
+432,2.9080519676208496,
+433,3.252742290496826,
+434,3.7637205123901367,
+435,4.262693881988525,
+436,3.373936176300049,
+437,2.6735756397247314,
+438,4.155456066131592,
+439,2.1575229167938232,
+440,3.4645214080810547,
+441,4.160406112670898,
+442,3.847316026687622,
+443,3.058614730834961,
+444,3.3499584197998047,
+445,3.798222303390503,
+446,3.033684730529785,
+447,3.512378215789795,
+448,2.828388214111328,
+449,4.068789482116699,
+450,4.078660011291504,3.412485858227344
+451,3.502532482147217,
+452,4.521447658538818,
+453,3.8204264640808105,
+454,3.152890205383301,
+455,2.990062713623047,
+456,4.356359481811523,
+457,4.176229953765869,
+458,4.333327770233154,

train_loss_curve.png

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d10485692c0d0e1d800972e4cb6c66ce6fa0f0945d90b10ec2233e17f3863b23
+size 235765