Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* 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

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
+}

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 +1,123 @@
+{
+  "architectures": [
+    "RizeForCausalLM"
+  ],
+  "attention_arch": "kimi_linear",
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration_rize.RizeConfig",
+    "AutoModel": "modeling_rize.RizeModel",
+    "AutoModelForCausalLM": "modeling_rize.RizeForCausalLM"
+  },
+  "aux_loss_alpha": 0.0,
+  "auxfree_bias_clip": 5.0,
+  "auxfree_bias_lr": 0.0,
+  "auxfree_update_interval": 8,
+  "block_diag_causal_mask": true,
+  "bos_token_id": null,
+  "collect_moe_stats": true,
+  "doc_sep_token": null,
+  "doc_sep_token_id": null,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "ep_size": 1,
+  "first_k_dense_replace": 1,
+  "force_chunk_attention": true,
+  "force_chunk_mode": true,
+  "force_training_chunk_mode": true,
+  "freeze_router_on_sft": true,
+  "global_lbl_buffer_across_ga": true,
+  "global_lbl_enabled": true,
+  "global_lbl_sync_across_ranks": true,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8192,
+  "kv_lora_rank": 512,
+  "linear_attn_config": {
+    "full_attn_layers": [
+      4,
+      8,
+      12,
+      16,
+      19
+    ],
+    "head_dim": 128,
+    "kda_layers": [
+      1,
+      2,
+      3,
+      5,
+      6,
+      7,
+      9,
+      10,
+      11,
+      13,
+      14,
+      15,
+      17,
+      18
+    ],
+    "num_heads": 12,
+    "short_conv_kernel_size": 4
+  },
+  "linear_ce_impl": "cce_exact",
+  "max_position_embeddings": 32768,
+  "model_type": "rize",
+  "moe_intermediate_size": 704,
+  "moe_layer_freq": 1,
+  "moe_router_active_only": true,
+  "n_group": 1,
+  "n_routed_experts": 64,
+  "n_shared_experts": 1,
+  "norm_topk_prob": true,
+  "num_attention_heads": 12,
+  "num_experts_per_tok": 4,
+  "num_hidden_layers": 19,
+  "num_key_value_heads": 12,
+  "num_nextn_predict_layers": 0,
+  "pad_token_id": 151643,
+  "pretraining_tp": 1,
+  "prompt_loss_weight": 0.1,
+  "q_lora_rank": null,
+  "qk_nope_head_dim": 128,
+  "qk_rope_head_dim": 64,
+  "reset_position_ids_per_sample": true,
+  "return_logits_in_train": false,
+  "rms_norm_eps": 1e-05,
+  "rope_beta_fast": 32.0,
+  "rope_beta_slow": 1.0,
+  "rope_mscale": 1.0,
+  "rope_mscale_all_dim": 1.0,
+  "rope_original_max_position_embeddings": 8192,
+  "rope_scaling": {
+    "beta_fast": 32.0,
+    "beta_slow": 1.0,
+    "factor": 4.0,
+    "mscale": 1.0,
+    "mscale_all_dim": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "yarn",
+    "type": "yarn"
+  },
+  "rope_scaling_factor": 4.0,
+  "rope_scaling_type": "yarn",
+  "rope_theta": 50000.0,
+  "routed_scaling_factor": 1.9984212625219382,
+  "sample_sep_is_eos": true,
+  "sample_sep_token": "<|im_end|>",
+  "sample_sep_token_id": 151645,
+  "scoring_func": "sigmoid",
+  "seq_aux": false,
+  "tie_word_embeddings": false,
+  "topk_group": 1,
+  "topk_method": "noaux_tc",
+  "transformers_version": "4.57.3",
+  "use_cache": false,
+  "use_doc_sep": false,
+  "use_linear_ce": true,
+  "v_head_dim": 128,
+  "vocab_size": 151669
+}

configuration_rize.py

@@ -0,0 +1,344 @@
+# Adapted from an upstream configuration file.
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+RIZE_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
+class RizeConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`RizeModel`]. It is used to instantiate an Rize
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the Rize-V3.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 129280):
+            Vocabulary size of the Rize model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`RizeModel`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 11008):
+            Dimension of the MLP representations.
+        moe_intermediate_size (`int`, *optional*, defaults to 1407):
+            Dimension of the MoE representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer decoder.
+        num_nextn_predict_layers (`int`, *optional*, defaults to 1):
+            Number of nextn predict layers in the RizeV3 Model.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer decoder.
+        n_shared_experts (`int`, *optional*, defaults to None):
+            Number of shared experts, None means dense model.
+        n_routed_experts (`int`, *optional*, defaults to None):
+            Number of routed experts, None means dense model.
+        routed_scaling_factor (`float`, *optional*, defaults to 1.0):
+            Scaling factor or routed experts.
+        topk_method (`str`, *optional*, defaults to `gready`):
+            Topk method used in routed gate.
+        n_group (`int`, *optional*, defaults to None):
+            Number of groups for routed experts.
+        topk_group (`int`, *optional*, defaults to None):
+            Number of selected groups for each token(for each token, ensuring the selected experts is only within `topk_group` groups).
+        num_experts_per_tok (`int`, *optional*, defaults to None):
+            Number of selected experts, None means dense model.
+        moe_layer_freq (`int`, *optional*, defaults to 1):
+            The frequency of the MoE layer: one expert layer for every `moe_layer_freq - 1` dense layers.
+        first_k_dense_replace (`int`, *optional*, defaults to 0):
+            Number of dense layers in shallow layers(embed->dense->dense->...->dense->moe->moe...->lm_head).
+                                                            \--k dense layers--/
+        norm_topk_prob (`bool`, *optional*, defaults to False):
+            Whether to normalize the weights of the routed experts.
+        scoring_func (`str`, *optional*, defaults to 'softmax'):
+            Method of computing expert weights.
+        aux_loss_alpha (`float`, *optional*, defaults to 0.001):
+            Auxiliary loss weight coefficient.
+        seq_aux = (`bool`, *optional*, defaults to True):
+            Whether to compute the auxiliary loss for each individual sample.
+        num_key_value_heads (`int`, *optional*):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
+            `num_attention_heads`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 2048):
+            The maximum sequence length that this model might ever be used with.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        pad_token_id (`int`, *optional*):
+            Padding token id.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            Beginning of stream token id.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            End of stream token id.
+        pretraining_tp (`int`, *optional*, defaults to 1):
+            Experimental feature. Tensor parallelism rank used during pretraining. Please refer to [this
+            document](https://huggingface.co/docs/transformers/parallelism) to understand more about it. This value is
+            necessary to ensure exact reproducibility of the pretraining results. Please refer to [this
+            issue](https://github.com/pytorch/pytorch/issues/76232).
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        rope_scaling (`Dict`, *optional*):
+            Dictionary containing the scaling configuration for the RoPE embeddings. Currently supports two scaling
+            strategies: linear and dynamic. Their scaling factor must be a float greater than 1. The expected format is
+            `{"type": strategy name, "factor": scaling factor}`. When using this flag, don't update
+            `max_position_embeddings` to the expected new maximum.
+        attention_bias (`bool`, defaults to `False`, *optional*, defaults to `False`):
+            Whether to use a bias in the query, key, value and output projection layers during self-attention.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        use_linear_ce (`bool`, *optional*, defaults to `True`):
+            Whether to use cut-cross-entropy (linear CE) for causal LM loss computation when labels are provided.
+        linear_ce_impl (`str`, *optional*, defaults to `"cce_exact"`):
+            Implementation name passed to cut-cross-entropy when `use_linear_ce=True`.
+
+    ```python
+    >>> from transformers import RizeModel, RizeConfig
+
+    >>> # Initializing a Rize-V3 style configuration
+    >>> configuration = RizeConfig()
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "rize"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=129280,
+        hidden_size=7168,
+        intermediate_size=18432,
+        moe_intermediate_size = 2048,
+        num_hidden_layers=61,
+        num_nextn_predict_layers=1,
+        num_attention_heads=128,
+        num_key_value_heads=128,
+        n_shared_experts = 1,
+        n_routed_experts = 256,
+        ep_size = 1,
+        routed_scaling_factor = 2.5,
+        kv_lora_rank = 512,
+        q_lora_rank = 1536,
+        qk_rope_head_dim = 64,
+        v_head_dim = 128,
+        qk_nope_head_dim = 128,
+        topk_method = 'noaux_tc',
+        n_group = 8,
+        topk_group = 4,
+        num_experts_per_tok = 8,
+        moe_layer_freq = 1,
+        first_k_dense_replace = 3,
+        norm_topk_prob = True,
+        scoring_func = 'sigmoid',
+        aux_loss_alpha = 0.001,
+        seq_aux = True,
+        auxfree_bias_lr = 0.0,
+        hidden_act="silu",
+        max_position_embeddings=4096,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=0,
+        eos_token_id=1,
+        pretraining_tp=1,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        rope_scaling=None,
+        attention_bias=False,
+        attention_dropout=0.0,
+        use_linear_ce=True,
+        linear_ce_impl="cce_exact",
+        attention_arch="auto",
+        linear_attn_config=None,
+        block_diag_causal_mask=False,
+        reset_position_ids_per_sample=False,
+        moe_router_active_only=True,
+        prompt_loss_weight=0.0,
+        freeze_router_on_sft=False,
+        global_lbl_enabled=False,
+        global_lbl_sync_across_ranks=False,
+        global_lbl_buffer_across_ga=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.moe_intermediate_size = moe_intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_nextn_predict_layers = num_nextn_predict_layers
+        self.num_attention_heads = num_attention_heads
+        self.n_shared_experts = n_shared_experts
+        self.n_routed_experts = n_routed_experts
+        self.ep_size = ep_size
+        self.routed_scaling_factor = routed_scaling_factor
+        self.kv_lora_rank = kv_lora_rank
+        self.q_lora_rank = q_lora_rank
+        self.qk_rope_head_dim = qk_rope_head_dim
+        self.v_head_dim = v_head_dim
+        self.qk_nope_head_dim = qk_nope_head_dim
+        self.topk_method = topk_method
+        self.n_group = n_group
+        self.topk_group = topk_group
+        self.num_experts_per_tok = num_experts_per_tok
+        self.moe_layer_freq = moe_layer_freq
+        self.first_k_dense_replace = first_k_dense_replace
+        self.norm_topk_prob = norm_topk_prob
+        self.scoring_func = scoring_func
+        self.aux_loss_alpha = aux_loss_alpha
+        self.seq_aux = seq_aux
+        self.auxfree_bias_lr = auxfree_bias_lr
+        # for backward compatibility
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.pretraining_tp = pretraining_tp
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.rope_scaling = rope_scaling
+        self.attention_bias = attention_bias
+        self.attention_dropout = attention_dropout
+        self.use_linear_ce = use_linear_ce
+        self.linear_ce_impl = linear_ce_impl
+        self.block_diag_causal_mask = bool(block_diag_causal_mask)
+        self.reset_position_ids_per_sample = bool(reset_position_ids_per_sample)
+        self.moe_router_active_only = bool(moe_router_active_only)
+        self.prompt_loss_weight = float(prompt_loss_weight)
+        self.freeze_router_on_sft = bool(freeze_router_on_sft)
+        self.global_lbl_enabled = bool(global_lbl_enabled)
+        self.global_lbl_sync_across_ranks = bool(global_lbl_sync_across_ranks)
+        self.global_lbl_buffer_across_ga = bool(global_lbl_buffer_across_ga)
+
+        # ---- Kimi-Linear / hybrid attention knobs (optional) ----
+        # attention_arch: 'auto' (default; infer from linear_attn_config), 'standard', or 'kimi_linear'
+        self.attention_arch = attention_arch
+        # linear_attn_config: dict or None
+        self.linear_attn_config = linear_attn_config
+
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+    @property
+    def is_linear_attn(self) -> bool:
+        """Whether this config enables the Kimi-Linear style hybrid attention.
+
+        Notes:
+          - This is intentionally a *property* (not a stored boolean) so that it works
+            for both new configs (attention_arch) and older checkpoints that might only
+            carry linear_attn_config.
+        """
+        arch = getattr(self, "attention_arch", None)
+        la_cfg = getattr(self, "linear_attn_config", None)
+        if arch is None:
+            return la_cfg is not None
+        # allow direct boolean override (e.g. programmatic configs)
+        if isinstance(arch, bool):
+            return bool(arch) and la_cfg is not None
+
+        arch_s = str(arch).lower().strip()
+        if arch_s in ("auto", "infer"):
+            return la_cfg is not None
+        if arch_s in ("standard", "full", "mla", "default", "none", ""):
+            return False
+        if arch_s in ("kimi_linear", "kda", "linear_attn", "linear", "hybrid"):
+            return la_cfg is not None
+
+        # Unknown value: be conservative (disable)
+        return False
+
+    @is_linear_attn.setter
+    def is_linear_attn(self, value: bool) -> None:
+        """Back-compat setter. Allows `config.is_linear_attn = True/False`."""
+        if bool(value):
+            # Only flip attention_arch if it is currently unset or standard-ish.
+            cur = getattr(self, "attention_arch", "standard")
+            if cur is None or str(cur).lower().strip() in ("standard", "full", "default", "none", ""):
+                self.attention_arch = "kimi_linear"
+        else:
+            self.attention_arch = "standard"
+
+    def is_kda_layer(self, layer_idx: int) -> bool:
+        """Return True if `layer_idx` (0-indexed) should use KDA (linear attention).
+
+        Layer index convention:
+          - Internally we expect `layer_idx` to be 0-indexed (as used by `range(num_hidden_layers)`).
+          - In config.linear_attn_config, `kda_layers` / `full_attn_layers` may be either:
+              * 0-indexed (0..num_hidden_layers-1), OR
+              * 1-indexed (1..num_hidden_layers) like the official Kimi-Linear configs.
+            We auto-detect the convention.
+        """
+        if not self.is_linear_attn:
+            return False
+
+        cfg = getattr(self, "linear_attn_config", None)
+        if not isinstance(cfg, dict):
+            return False
+
+        if layer_idx is None:
+            return False
+        try:
+            layer_idx = int(layer_idx)
+        except Exception:
+            return False
+        if layer_idx < 0 or layer_idx >= int(getattr(self, "num_hidden_layers", 0) or 0):
+            return False
+
+        kda_layers = cfg.get("kda_layers", None)
+        full_layers = cfg.get("full_attn_layers", None)
+
+        # Normalize lists (best-effort)
+        kda = [int(x) for x in (kda_layers or [])] if isinstance(kda_layers, (list, tuple)) else []
+        full = [int(x) for x in (full_layers or [])] if isinstance(full_layers, (list, tuple)) else []
+
+        # Auto-detect index base for the config lists
+        all_idx = kda + full
+        one_indexed = False
+        if all_idx:
+            n = int(getattr(self, "num_hidden_layers", 0) or 0)
+            # Strong signals:
+            if 0 in all_idx:
+                one_indexed = False
+            elif n in all_idx:
+                one_indexed = True
+            else:
+                # Heuristic: if everything is within [1, n], treat as 1-indexed
+                mn, mx = min(all_idx), max(all_idx)
+                if mn >= 1 and mx <= n:
+                    one_indexed = True
+
+        query_idx = layer_idx + (1 if one_indexed else 0)
+
+        if kda:
+            return query_idx in set(kda)
+        if full:
+            return query_idx not in set(full)
+
+        # If neither list is provided, default to "no KDA"
+        return False

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": [
+    151645,
+    163585
+  ],
+  "pad_token_id": 151643,
+  "transformers_version": "4.57.3"
+}

model-00001-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80982e117b7681afea7bc763935b39916e9e8fa5ac9f74f615ec94016c2fdc1a
+size 5000146488

model-00002-of-00002.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcd5c4e76e857e82faf5c744a6353bd938ceadd0a692c58941a11a2a34e3c480
+size 3973275080

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": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.json

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

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": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

trainer_state.json

@@ -0,0 +1,883 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03333333333333333,
+      "grad_norm": 10.375,
+      "learning_rate": 1.8e-07,
+      "loss": 2.1128,
+      "slime/anchor_loss": 0.5652838359801535,
+      "slime/chosen_len": 1131.75390625,
+      "slime/chosen_logp": -1.2561863364077,
+      "slime/chosen_sft_loss": 1.2561863364077,
+      "slime/chosen_sft_weighted_loss": 0.025123726156539305,
+      "slime/dist_loss": 0.018970487654221914,
+      "slime/hard_satisfied": 0.33203125,
+      "slime/loss": 2.1128057595517022,
+      "slime/margin": 0.19336955933831632,
+      "slime/rejected_len": 2604.171875,
+      "slime/rejected_logp": -1.449555895573576,
+      "slime/rejected_stabilizer": 7.517138462670846,
+      "slime/soft_gate": 0.4965854723122902,
+      "slime/stabilize_loss": 1.5034277173785084,
+      "source_mix/carefully": 0.1796875,
+      "source_mix/dolci-inst": 0.13046875,
+      "source_mix/dolci-think": 0.128125,
+      "source_mix/japanese": 0.134375,
+      "source_mix/math": 0.140625,
+      "source_mix/nvidia-cascade": 0.25078125,
+      "source_mix/self-intro": 0.0359375,
+      "step": 10
+    },
+    {
+      "epoch": 0.06666666666666667,
+      "grad_norm": 270.0,
+      "learning_rate": 3.7999999999999996e-07,
+      "loss": 2.0585,
+      "slime/anchor_loss": 0.5715644719993861,
+      "slime/chosen_len": 1118.2375,
+      "slime/chosen_logp": -1.2701433055754705,
+      "slime/chosen_sft_loss": 1.2701433055754705,
+      "slime/chosen_sft_weighted_loss": 0.025402865548713117,
+      "slime/dist_loss": 0.018322167105907283,
+      "slime/hard_satisfied": 0.3109375,
+      "slime/loss": 2.0585288689151637,
+      "slime/margin": 0.1740653719578404,
+      "slime/rejected_len": 2716.92578125,
+      "slime/rejected_logp": -1.4442086774139853,
+      "slime/rejected_stabilizer": 7.2161967355583325,
+      "slime/soft_gate": 0.4972848696401343,
+      "slime/stabilize_loss": 1.4432393665541894,
+      "source_mix/carefully": 0.171875,
+      "source_mix/dolci-inst": 0.1328125,
+      "source_mix/dolci-think": 0.11875,
+      "source_mix/japanese": 0.12421875,
+      "source_mix/math": 0.13515625,
+      "source_mix/nvidia-cascade": 0.275,
+      "source_mix/self-intro": 0.0421875,
+      "step": 20
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 64.5,
+      "learning_rate": 5.8e-07,
+      "loss": 2.0348,
+      "slime/anchor_loss": 0.5777066876407844,
+      "slime/chosen_len": 1126.65390625,
+      "slime/chosen_logp": -1.283792673991411,
+      "slime/chosen_sft_loss": 1.283792673991411,
+      "slime/chosen_sft_weighted_loss": 0.025675852919619047,
+      "slime/dist_loss": 0.01966060851878524,
+      "slime/hard_satisfied": 0.30078125,
+      "slime/loss": 2.0347598268243017,
+      "slime/margin": 0.16343559486558662,
+      "slime/rejected_len": 2952.1328125,
+      "slime/rejected_logp": -1.4472282689312124,
+      "slime/rejected_stabilizer": 7.05858330585761,
+      "slime/soft_gate": 0.49771429020911456,
+      "slime/stabilize_loss": 1.4117166813746735,
+      "source_mix/carefully": 0.17421875,
+      "source_mix/dolci-inst": 0.13203125,
+      "source_mix/dolci-think": 0.12421875,
+      "source_mix/japanese": 0.12734375,
+      "source_mix/math": 0.14453125,
+      "source_mix/nvidia-cascade": 0.24921875,
+      "source_mix/self-intro": 0.0484375,
+      "step": 30
+    },
+    {
+      "epoch": 0.13333333333333333,
+      "grad_norm": 13.5,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 2.1862,
+      "slime/anchor_loss": 0.5970213656022679,
+      "slime/chosen_len": 1085.8375,
+      "slime/chosen_logp": -1.3267141812248155,
+      "slime/chosen_sft_loss": 1.3267141812248155,
+      "slime/chosen_sft_weighted_loss": 0.026534283025539195,
+      "slime/dist_loss": 0.019207629275163285,
+      "slime/hard_satisfied": 0.29609375,
+      "slime/loss": 2.1861922092386523,
+      "slime/margin": 0.17782343912986107,
+      "slime/rejected_len": 2872.32890625,
+      "slime/rejected_logp": -1.5045376207468508,
+      "slime/rejected_stabilizer": 7.717144486965845,
+      "slime/soft_gate": 0.49718887559138236,
+      "slime/stabilize_loss": 1.5434289268159773,
+      "source_mix/carefully": 0.18203125,
+      "source_mix/dolci-inst": 0.115625,
+      "source_mix/dolci-think": 0.12734375,
+      "source_mix/japanese": 0.13203125,
+      "source_mix/math": 0.1328125,
+      "source_mix/nvidia-cascade": 0.26328125,
+      "source_mix/self-intro": 0.046875,
+      "step": 40
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 34.5,
+      "learning_rate": 9.8e-07,
+      "loss": 1.9821,
+      "slime/anchor_loss": 0.5981764935819228,
+      "slime/chosen_len": 1107.9546875,
+      "slime/chosen_logp": -1.3292811316568987,
+      "slime/chosen_sft_loss": 1.3292811316568987,
+      "slime/chosen_sft_weighted_loss": 0.026585622042512115,
+      "slime/dist_loss": 0.02136569871681928,
+      "slime/hard_satisfied": 0.25234375,
+      "slime/loss": 1.9820999772753567,
+      "slime/margin": 0.07440594642976066,
+      "slime/rejected_len": 2603.40625,
+      "slime/rejected_logp": -1.4036870779033053,
+      "slime/rejected_stabilizer": 6.679860710317735,
+      "slime/soft_gate": 0.5009892466012389,
+      "slime/stabilize_loss": 1.3359721658394847,
+      "source_mix/carefully": 0.178125,
+      "source_mix/dolci-inst": 0.11875,
+      "source_mix/dolci-think": 0.13359375,
+      "source_mix/japanese": 0.12578125,
+      "source_mix/math": 0.1265625,
+      "source_mix/nvidia-cascade": 0.275,
+      "source_mix/self-intro": 0.0421875,
+      "step": 50
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 20.875,
+      "learning_rate": 1.18e-06,
+      "loss": 2.0138,
+      "slime/anchor_loss": 0.5834041117399466,
+      "slime/chosen_len": 1091.12109375,
+      "slime/chosen_logp": -1.2964536144820158,
+      "slime/chosen_sft_loss": 1.2964536144820158,
+      "slime/chosen_sft_weighted_loss": 0.025929071675750492,
+      "slime/dist_loss": 0.02069701280749996,
+      "slime/hard_satisfied": 0.28046875,
+      "slime/loss": 2.0138427876867353,
+      "slime/margin": 0.12427767315530218,
+      "slime/rejected_len": 2604.715625,
+      "slime/rejected_logp": -1.4207312876413198,
+      "slime/rejected_stabilizer": 6.91906284771976,
+      "slime/soft_gate": 0.4991721922764555,
+      "slime/stabilize_loss": 1.3838125903435867,
+      "source_mix/carefully": 0.1578125,
+      "source_mix/dolci-inst": 0.14609375,
+      "source_mix/dolci-think": 0.11171875,
+      "source_mix/japanese": 0.115625,
+      "source_mix/math": 0.13125,
+      "source_mix/nvidia-cascade": 0.28984375,
+      "source_mix/self-intro": 0.04765625,
+      "step": 60
+    },
+    {
+      "epoch": 0.23333333333333334,
+      "grad_norm": 6.09375,
+      "learning_rate": 1.38e-06,
+      "loss": 2.093,
+      "slime/anchor_loss": 0.5975191399382311,
+      "slime/chosen_len": 1169.30859375,
+      "slime/chosen_logp": -1.3278203454508912,
+      "slime/chosen_sft_loss": 1.3278203454508912,
+      "slime/chosen_sft_weighted_loss": 0.02655640635189229,
+      "slime/dist_loss": 0.02211085479052599,
+      "slime/hard_satisfied": 0.28359375,
+      "slime/loss": 2.0930444569152313,
+      "slime/margin": 0.11257743685273454,
+      "slime/rejected_len": 2785.41328125,
+      "slime/rejected_logp": -1.4403977824331378,
+      "slime/rejected_stabilizer": 7.234290163306286,
+      "slime/soft_gate": 0.49962827693670986,
+      "slime/stabilize_loss": 1.4468580567918252,
+      "source_mix/carefully": 0.17890625,
+      "source_mix/dolci-inst": 0.1203125,
+      "source_mix/dolci-think": 0.11875,
+      "source_mix/japanese": 0.140625,
+      "source_mix/math": 0.134375,
+      "source_mix/nvidia-cascade": 0.2640625,
+      "source_mix/self-intro": 0.04296875,
+      "step": 70
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 84.5,
+      "learning_rate": 1.58e-06,
+      "loss": 1.938,
+      "slime/anchor_loss": 0.5820682898134691,
+      "slime/chosen_len": 1132.0359375,
+      "slime/chosen_logp": -1.2934851201134734,
+      "slime/chosen_sft_loss": 1.2934851201134734,
+      "slime/chosen_sft_weighted_loss": 0.02586970183710946,
+      "slime/dist_loss": 0.02047131769329553,
+      "slime/hard_satisfied": 0.28671875,
+      "slime/loss": 1.9379910714080324,
+      "slime/margin": 0.10804048473073635,
+      "slime/rejected_len": 2687.40234375,
+      "slime/rejected_logp": -1.4015256044144109,
+      "slime/rejected_stabilizer": 6.547908684506547,
+      "slime/soft_gate": 0.499736803304404,
+      "slime/stabilize_loss": 1.3095817611982057,
+      "source_mix/carefully": 0.17734375,
+      "source_mix/dolci-inst": 0.13125,
+      "source_mix/dolci-think": 0.121875,
+      "source_mix/japanese": 0.12890625,
+      "source_mix/math": 0.13203125,
+      "source_mix/nvidia-cascade": 0.26015625,
+      "source_mix/self-intro": 0.0484375,
+      "step": 80
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 32.5,
+      "learning_rate": 1.78e-06,
+      "loss": 1.9334,
+      "slime/anchor_loss": 0.585230390515062,
+      "slime/chosen_len": 1031.903125,
+      "slime/chosen_logp": -1.300512012524996,
+      "slime/chosen_sft_loss": 1.300512012524996,
+      "slime/chosen_sft_weighted_loss": 0.026010239717879812,
+      "slime/dist_loss": 0.020852691849732707,
+      "slime/hard_satisfied": 0.28203125,
+      "slime/loss": 1.9333723544841632,
+      "slime/margin": 0.10233128607505933,
+      "slime/rejected_len": 2716.9671875,
+      "slime/rejected_logp": -1.4028432980827348,
+      "slime/rejected_stabilizer": 6.5063950447482055,
+      "slime/soft_gate": 0.499991340667475,
+      "slime/stabilize_loss": 1.301279031730519,
+      "source_mix/carefully": 0.165625,
+      "source_mix/dolci-inst": 0.128125,
+      "source_mix/dolci-think": 0.12578125,
+      "source_mix/japanese": 0.13203125,
+      "source_mix/math": 0.12734375,
+      "source_mix/nvidia-cascade": 0.27265625,
+      "source_mix/self-intro": 0.0484375,
+      "step": 90
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 17.125,
+      "learning_rate": 1.98e-06,
+      "loss": 2.0311,
+      "slime/anchor_loss": 0.5909910393456812,
+      "slime/chosen_len": 1149.38828125,
+      "slime/chosen_logp": -1.3133134559844621,
+      "slime/chosen_sft_loss": 1.3133134559844621,
+      "slime/chosen_sft_weighted_loss": 0.026266268511881206,
+      "slime/dist_loss": 0.020266785123857288,
+      "slime/hard_satisfied": 0.28125,
+      "slime/loss": 2.0310499653685836,
+      "slime/margin": 0.12902917331666686,
+      "slime/rejected_len": 2606.6953125,
+      "slime/rejected_logp": -1.4423426295616082,
+      "slime/rejected_stabilizer": 6.967629258579109,
+      "slime/soft_gate": 0.4989678652025759,
+      "slime/stabilize_loss": 1.393525873884937,
+      "source_mix/carefully": 0.17109375,
+      "source_mix/dolci-inst": 0.12265625,
+      "source_mix/dolci-think": 0.13046875,
+      "source_mix/japanese": 0.12421875,
+      "source_mix/math": 0.13515625,
+      "source_mix/nvidia-cascade": 0.26484375,
+      "source_mix/self-intro": 0.0515625,
+      "step": 100
+    },
+    {
+      "epoch": 0.36666666666666664,
+      "grad_norm": 14.1875,
+      "learning_rate": 2e-06,
+      "loss": 1.8712,
+      "slime/anchor_loss": 0.5815936719751335,
+      "slime/chosen_len": 1120.3484375,
+      "slime/chosen_logp": -1.2924304147367365,
+      "slime/chosen_sft_loss": 1.2924304147367365,
+      "slime/chosen_sft_weighted_loss": 0.02584860770148225,
+      "slime/dist_loss": 0.0211684020853113,
+      "slime/hard_satisfied": 0.26875,
+      "slime/loss": 1.8711517222691327,
+      "slime/margin": 0.08797086402191781,
+      "slime/rejected_len": 2819.31171875,
+      "slime/rejected_logp": -1.3804012785600208,
+      "slime/rejected_stabilizer": 6.21270509515889,
+      "slime/soft_gate": 0.5005081544164568,
+      "slime/stabilize_loss": 1.2425410391850165,
+      "source_mix/carefully": 0.16875,
+      "source_mix/dolci-inst": 0.12734375,
+      "source_mix/dolci-think": 0.12421875,
+      "source_mix/japanese": 0.13515625,
+      "source_mix/math": 0.13125,
+      "source_mix/nvidia-cascade": 0.25625,
+      "source_mix/self-intro": 0.05703125,
+      "step": 110
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 512.0,
+      "learning_rate": 2e-06,
+      "loss": 1.8067,
+      "slime/anchor_loss": 0.5661823063972407,
+      "slime/chosen_len": 1122.00234375,
+      "slime/chosen_logp": -1.2581829352478962,
+      "slime/chosen_sft_loss": 1.2581829352478962,
+      "slime/chosen_sft_weighted_loss": 0.02516365816418329,
+      "slime/dist_loss": 0.01988503187535571,
+      "slime/hard_satisfied": 0.27734375,
+      "slime/loss": 1.8066898226796184,
+      "slime/margin": 0.10299149260972626,
+      "slime/rejected_len": 2811.6,
+      "slime/rejected_logp": -1.361174427827791,
+      "slime/rejected_stabilizer": 5.977294027025346,
+      "slime/soft_gate": 0.4999435399658978,
+      "slime/stabilize_loss": 1.1954588271291868,
+      "source_mix/carefully": 0.1640625,
+      "source_mix/dolci-inst": 0.13125,
+      "source_mix/dolci-think": 0.1375,
+      "source_mix/japanese": 0.13515625,
+      "source_mix/math": 0.12734375,
+      "source_mix/nvidia-cascade": 0.259375,
+      "source_mix/self-intro": 0.0453125,
+      "step": 120
+    },
+    {
+      "epoch": 0.43333333333333335,
+      "grad_norm": 64.0,
+      "learning_rate": 2e-06,
+      "loss": 1.8196,
+      "slime/anchor_loss": 0.5704214560333639,
+      "slime/chosen_len": 1151.52578125,
+      "slime/chosen_logp": -1.2676032666116952,
+      "slime/chosen_sft_loss": 1.2676032666116952,
+      "slime/chosen_sft_weighted_loss": 0.02535206476932217,
+      "slime/dist_loss": 0.02142226323093155,
+      "slime/hard_satisfied": 0.24765625,
+      "slime/loss": 1.8195731153886299,
+      "slime/margin": 0.07547286066692323,
+      "slime/rejected_len": 2794.225,
+      "slime/rejected_logp": -1.3430761275914846,
+      "slime/rejected_stabilizer": 6.011886543076253,
+      "slime/soft_gate": 0.5010092506767251,
+      "slime/stabilize_loss": 1.2023773317247106,
+      "source_mix/carefully": 0.16015625,
+      "source_mix/dolci-inst": 0.13671875,
+      "source_mix/dolci-think": 0.125,
+      "source_mix/japanese": 0.14296875,
+      "source_mix/math": 0.11484375,
+      "source_mix/nvidia-cascade": 0.26640625,
+      "source_mix/self-intro": 0.05390625,
+      "step": 130
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 18.625,
+      "learning_rate": 2e-06,
+      "loss": 1.7939,
+      "slime/anchor_loss": 0.5643214548053948,
+      "slime/chosen_len": 1126.16015625,
+      "slime/chosen_logp": -1.2540477099530107,
+      "slime/chosen_sft_loss": 1.2540477099530107,
+      "slime/chosen_sft_weighted_loss": 0.02508095362707081,
+      "slime/dist_loss": 0.020207823722915918,
+      "slime/hard_satisfied": 0.284375,
+      "slime/loss": 1.793889444656088,
+      "slime/margin": 0.11795559519378003,
+      "slime/rejected_len": 2949.6734375,
+      "slime/rejected_logp": -1.3720033054185479,
+      "slime/rejected_stabilizer": 5.92139595746994,
+      "slime/soft_gate": 0.49938542693853377,
+      "slime/stabilize_loss": 1.1842792140545497,
+      "source_mix/carefully": 0.19140625,
+      "source_mix/dolci-inst": 0.14140625,
+      "source_mix/dolci-think": 0.1125,
+      "source_mix/japanese": 0.14140625,
+      "source_mix/math": 0.1359375,
+      "source_mix/nvidia-cascade": 0.240625,
+      "source_mix/self-intro": 0.03671875,
+      "step": 140
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 50.75,
+      "learning_rate": 2e-06,
+      "loss": 1.742,
+      "slime/anchor_loss": 0.5520623270414944,
+      "slime/chosen_len": 1227.03125,
+      "slime/chosen_logp": -1.2268052033527055,
+      "slime/chosen_sft_loss": 1.2268052033527055,
+      "slime/chosen_sft_weighted_loss": 0.02453610350275994,
+      "slime/dist_loss": 0.01930630389658745,
+      "slime/hard_satisfied": 0.28203125,
+      "slime/loss": 1.742006646405207,
+      "slime/margin": 0.12888034964853431,
+      "slime/rejected_len": 2967.546875,
+      "slime/rejected_logp": -1.3556855528629967,
+      "slime/rejected_stabilizer": 5.7305094635114076,
+      "slime/soft_gate": 0.4989769039209932,
+      "slime/stabilize_loss": 1.1461019126836618,
+      "source_mix/carefully": 0.16796875,
+      "source_mix/dolci-inst": 0.12578125,
+      "source_mix/dolci-think": 0.14140625,
+      "source_mix/japanese": 0.13125,
+      "source_mix/math": 0.12890625,
+      "source_mix/nvidia-cascade": 0.2578125,
+      "source_mix/self-intro": 0.046875,
+      "step": 150
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 43.0,
+      "learning_rate": 2e-06,
+      "loss": 1.7376,
+      "slime/anchor_loss": 0.5569766616019478,
+      "slime/chosen_len": 1184.3328125,
+      "slime/chosen_logp": -1.2377259450149722,
+      "slime/chosen_sft_loss": 1.2377259450149722,
+      "slime/chosen_sft_weighted_loss": 0.02475451834600335,
+      "slime/dist_loss": 0.020114551604763165,
+      "slime/hard_satisfied": 0.28203125,
+      "slime/loss": 1.7375932445982472,
+      "slime/margin": 0.11652939966297708,
+      "slime/rejected_len": 2700.2,
+      "slime/rejected_logp": -1.354255344721969,
+      "slime/rejected_stabilizer": 5.678737456933595,
+      "slime/soft_gate": 0.49942631730809806,
+      "slime/stabilize_loss": 1.1357475130011152,
+      "source_mix/carefully": 0.178125,
+      "source_mix/dolci-inst": 0.1234375,
+      "source_mix/dolci-think": 0.13125,
+      "source_mix/japanese": 0.134375,
+      "source_mix/math": 0.1234375,
+      "source_mix/nvidia-cascade": 0.2640625,
+      "source_mix/self-intro": 0.0453125,
+      "step": 160
+    },
+    {
+      "epoch": 0.5666666666666667,
+      "grad_norm": 19.75,
+      "learning_rate": 2e-06,
+      "loss": 1.5782,
+      "slime/anchor_loss": 0.532384615248884,
+      "slime/chosen_len": 1129.96015625,
+      "slime/chosen_logp": -1.183076953816635,
+      "slime/chosen_sft_loss": 1.183076953816635,
+      "slime/chosen_sft_weighted_loss": 0.02366153856519304,
+      "slime/dist_loss": 0.019202208291005718,
+      "slime/hard_satisfied": 0.28671875,
+      "slime/loss": 1.5782210501580267,
+      "slime/margin": 0.11024855870637111,
+      "slime/rejected_len": 3042.89375,
+      "slime/rejected_logp": -1.2933255127398298,
+      "slime/rejected_stabilizer": 5.014863350742962,
+      "slime/soft_gate": 0.4996744911884889,
+      "slime/stabilize_loss": 1.0029726890483288,
+      "source_mix/carefully": 0.159375,
+      "source_mix/dolci-inst": 0.12421875,
+      "source_mix/dolci-think": 0.13359375,
+      "source_mix/japanese": 0.13671875,
+      "source_mix/math": 0.1546875,
+      "source_mix/nvidia-cascade": 0.26171875,
+      "source_mix/self-intro": 0.0296875,
+      "step": 170
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.15625,
+      "learning_rate": 2e-06,
+      "loss": 1.4583,
+      "slime/anchor_loss": 0.5199616495083319,
+      "slime/chosen_len": 1213.884375,
+      "slime/chosen_logp": -1.155470362820779,
+      "slime/chosen_sft_loss": 1.155470362820779,
+      "slime/chosen_sft_weighted_loss": 0.023109406742059947,
+      "slime/dist_loss": 0.01785762085306395,
+      "slime/hard_satisfied": 0.2859375,
+      "slime/loss": 1.4582647689734585,
+      "slime/margin": 0.10910382422734984,
+      "slime/rejected_len": 2908.109375,
+      "slime/rejected_logp": -1.2645741872747749,
+      "slime/rejected_stabilizer": 4.486680382909253,
+      "slime/soft_gate": 0.4996923903701827,
+      "slime/stabilize_loss": 0.8973360939584382,
+      "source_mix/carefully": 0.16484375,
+      "source_mix/dolci-inst": 0.1328125,
+      "source_mix/dolci-think": 0.14296875,
+      "source_mix/japanese": 0.128125,
+      "source_mix/math": 0.14765625,
+      "source_mix/nvidia-cascade": 0.23828125,
+      "source_mix/self-intro": 0.0453125,
+      "step": 180
+    },
+    {
+      "epoch": 0.6333333333333333,
+      "grad_norm": 7.0625,
+      "learning_rate": 2e-06,
+      "loss": 1.5619,
+      "slime/anchor_loss": 0.5451826707037981,
+      "slime/chosen_len": 1196.08359375,
+      "slime/chosen_logp": -1.2115170773729915,
+      "slime/chosen_sft_loss": 1.2115170773729915,
+      "slime/chosen_sft_weighted_loss": 0.02423034098987955,
+      "slime/dist_loss": 0.018773801988095327,
+      "slime/hard_satisfied": 0.2953125,
+      "slime/loss": 1.5619090026302729,
+      "slime/margin": 0.12265400608303026,
+      "slime/rejected_len": 2788.084375,
+      "slime/rejected_logp": -1.3341710836619314,
+      "slime/rejected_stabilizer": 4.868610856326995,
+      "slime/soft_gate": 0.4991808100603521,
+      "slime/stabilize_loss": 0.9737221888412023,
+      "source_mix/carefully": 0.2046875,
+      "source_mix/dolci-inst": 0.11875,
+      "source_mix/dolci-think": 0.1140625,
+      "source_mix/japanese": 0.13046875,
+      "source_mix/math": 0.1359375,
+      "source_mix/nvidia-cascade": 0.24765625,
+      "source_mix/self-intro": 0.0484375,
+      "step": 190
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 6.625,
+      "learning_rate": 2e-06,
+      "loss": 1.5499,
+      "slime/anchor_loss": 0.5411809546541917,
+      "slime/chosen_len": 1106.7875,
+      "slime/chosen_logp": -1.2026243753440213,
+      "slime/chosen_sft_loss": 1.2026243753440213,
+      "slime/chosen_sft_weighted_loss": 0.02405248696841227,
+      "slime/dist_loss": 0.01921966159346198,
+      "slime/hard_satisfied": 0.2859375,
+      "slime/loss": 1.5498943640850484,
+      "slime/margin": 0.11554879848263226,
+      "slime/rejected_len": 2395.91640625,
+      "slime/rejected_logp": -1.3181731744156422,
+      "slime/rejected_stabilizer": 4.827206226118141,
+      "slime/soft_gate": 0.49945375400129705,
+      "slime/stabilize_loss": 0.9654412612908345,
+      "source_mix/carefully": 0.19765625,
+      "source_mix/dolci-inst": 0.12578125,
+      "source_mix/dolci-think": 0.12734375,
+      "source_mix/japanese": 0.11953125,
+      "source_mix/math": 0.1203125,
+      "source_mix/nvidia-cascade": 0.27109375,
+      "source_mix/self-intro": 0.03828125,
+      "step": 200
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 44.75,
+      "learning_rate": 2e-06,
+      "loss": 1.4603,
+      "slime/anchor_loss": 0.529536109342007,
+      "slime/chosen_len": 1068.109375,
+      "slime/chosen_logp": -1.176746938750148,
+      "slime/chosen_sft_loss": 1.176746938750148,
+      "slime/chosen_sft_weighted_loss": 0.02353493825548867,
+      "slime/dist_loss": 0.01862544146329128,
+      "slime/hard_satisfied": 0.290625,
+      "slime/loss": 1.4602733735518996,
+      "slime/margin": 0.11214757621055468,
+      "slime/rejected_len": 2600.8359375,
+      "slime/rejected_logp": -1.288894514884305,
+      "slime/rejected_stabilizer": 4.442884352814872,
+      "slime/soft_gate": 0.49956530551426115,
+      "slime/stabilize_loss": 0.8885768864871352,
+      "source_mix/carefully": 0.1984375,
+      "source_mix/dolci-inst": 0.125,
+      "source_mix/dolci-think": 0.11328125,
+      "source_mix/japanese": 0.1203125,
+      "source_mix/math": 0.1328125,
+      "source_mix/nvidia-cascade": 0.259375,
+      "source_mix/self-intro": 0.05078125,
+      "step": 210
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 13.8125,
+      "learning_rate": 2e-06,
+      "loss": 1.3529,
+      "slime/anchor_loss": 0.5180962575308513,
+      "slime/chosen_len": 1267.4875,
+      "slime/chosen_logp": -1.1513250483461888,
+      "slime/chosen_sft_loss": 1.1513250483461888,
+      "slime/chosen_sft_weighted_loss": 0.023026500427795325,
+      "slime/dist_loss": 0.018737165465664417,
+      "slime/hard_satisfied": 0.271875,
+      "slime/loss": 1.3529100731015204,
+      "slime/margin": 0.10039320702198892,
+      "slime/rejected_len": 2910.2328125,
+      "slime/rejected_logp": -1.2517182552608574,
+      "slime/rejected_stabilizer": 3.965250689495588,
+      "slime/soft_gate": 0.5000133744208142,
+      "slime/stabilize_loss": 0.7930501511385956,
+      "source_mix/carefully": 0.18984375,
+      "source_mix/dolci-inst": 0.11015625,
+      "source_mix/dolci-think": 0.15390625,
+      "source_mix/japanese": 0.1375,
+      "source_mix/math": 0.13203125,
+      "source_mix/nvidia-cascade": 0.234375,
+      "source_mix/self-intro": 0.0421875,
+      "step": 220
+    },
+    {
+      "epoch": 0.7666666666666667,
+      "grad_norm": 15.375,
+      "learning_rate": 2e-06,
+      "loss": 1.3448,
+      "slime/anchor_loss": 0.5111413402909193,
+      "slime/chosen_len": 1182.5296875,
+      "slime/chosen_logp": -1.1358696755909476,
+      "slime/chosen_sft_loss": 1.1358696755909476,
+      "slime/chosen_sft_weighted_loss": 0.022717393007877007,
+      "slime/dist_loss": 0.017774005938222714,
+      "slime/hard_satisfied": 0.2984375,
+      "slime/loss": 1.344768432615092,
+      "slime/margin": 0.13050566065649036,
+      "slime/rejected_len": 2824.88515625,
+      "slime/rejected_logp": -1.2663753359134717,
+      "slime/rejected_stabilizer": 3.9656784189050085,
+      "slime/soft_gate": 0.49888017314951866,
+      "slime/stabilize_loss": 0.7931356976223469,
+      "source_mix/carefully": 0.175,
+      "source_mix/dolci-inst": 0.13203125,
+      "source_mix/dolci-think": 0.128125,
+      "source_mix/japanese": 0.11640625,
+      "source_mix/math": 0.134375,
+      "source_mix/nvidia-cascade": 0.26484375,
+      "source_mix/self-intro": 0.04921875,
+      "step": 230
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.6875,
+      "learning_rate": 2e-06,
+      "loss": 1.3167,
+      "slime/anchor_loss": 0.513396046102207,
+      "slime/chosen_len": 1151.9859375,
+      "slime/chosen_logp": -1.140880131906306,
+      "slime/chosen_sft_loss": 1.140880131906306,
+      "slime/chosen_sft_weighted_loss": 0.02281760216501425,
+      "slime/dist_loss": 0.01858289779211191,
+      "slime/hard_satisfied": 0.27578125,
+      "slime/loss": 1.316708378191106,
+      "slime/margin": 0.10129224676784361,
+      "slime/rejected_len": 2875.41640625,
+      "slime/rejected_logp": -1.242172378001851,
+      "slime/rejected_stabilizer": 3.8095590993121733,
+      "slime/soft_gate": 0.499989597639069,
+      "slime/stabilize_loss": 0.7619118337344843,
+      "source_mix/carefully": 0.1625,
+      "source_mix/dolci-inst": 0.11640625,
+      "source_mix/dolci-think": 0.13984375,
+      "source_mix/japanese": 0.14609375,
+      "source_mix/math": 0.125,
+      "source_mix/nvidia-cascade": 0.2625,
+      "source_mix/self-intro": 0.04765625,
+      "step": 240
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 17.125,
+      "learning_rate": 2e-06,
+      "loss": 1.2944,
+      "slime/anchor_loss": 0.5040076806962361,
+      "slime/chosen_len": 1084.5609375,
+      "slime/chosen_logp": -1.1200170961950788,
+      "slime/chosen_sft_loss": 1.1200170961950788,
+      "slime/chosen_sft_weighted_loss": 0.022400341415534087,
+      "slime/dist_loss": 0.017458863519634348,
+      "slime/hard_satisfied": 0.3078125,
+      "slime/loss": 1.2944442071835511,
+      "slime/margin": 0.12628607385559007,
+      "slime/rejected_len": 2848.92421875,
+      "slime/rejected_logp": -1.2463031702776788,
+      "slime/rejected_stabilizer": 3.752886550995754,
+      "slime/soft_gate": 0.4990315197966993,
+      "slime/stabilize_loss": 0.7505773224926088,
+      "source_mix/carefully": 0.171875,
+      "source_mix/dolci-inst": 0.1296875,
+      "source_mix/dolci-think": 0.13203125,
+      "source_mix/japanese": 0.1234375,
+      "source_mix/math": 0.13984375,
+      "source_mix/nvidia-cascade": 0.2546875,
+      "source_mix/self-intro": 0.0484375,
+      "step": 250
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 5.59375,
+      "learning_rate": 2e-06,
+      "loss": 1.286,
+      "slime/anchor_loss": 0.4994887421526073,
+      "slime/chosen_len": 1101.80625,
+      "slime/chosen_logp": -1.109975012994255,
+      "slime/chosen_sft_loss": 1.109975012994255,
+      "slime/chosen_sft_weighted_loss": 0.02219949978898512,
+      "slime/dist_loss": 0.017372727428653433,
+      "slime/hard_satisfied": 0.3171875,
+      "slime/loss": 1.2859536529314937,
+      "slime/margin": 0.1506339466868667,
+      "slime/rejected_len": 2640.659375,
+      "slime/rejected_logp": -1.2606089603868895,
+      "slime/rejected_stabilizer": 3.7344633417378645,
+      "slime/soft_gate": 0.4981274274876341,
+      "slime/stabilize_loss": 0.7468926805006049,
+      "source_mix/carefully": 0.16953125,
+      "source_mix/dolci-inst": 0.13359375,
+      "source_mix/dolci-think": 0.1171875,
+      "source_mix/japanese": 0.13828125,
+      "source_mix/math": 0.1375,
+      "source_mix/nvidia-cascade": 0.25625,
+      "source_mix/self-intro": 0.04765625,
+      "step": 260
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 13.375,
+      "learning_rate": 2e-06,
+      "loss": 1.272,
+      "slime/anchor_loss": 0.507745109476673,
+      "slime/chosen_len": 1099.45078125,
+      "slime/chosen_logp": -1.128322493477026,
+      "slime/chosen_sft_loss": 1.128322493477026,
+      "slime/chosen_sft_weighted_loss": 0.022566449361943342,
+      "slime/dist_loss": 0.018198056882033596,
+      "slime/hard_satisfied": 0.29921875,
+      "slime/loss": 1.271971960720839,
+      "slime/margin": 0.119626733957557,
+      "slime/rejected_len": 2296.753125,
+      "slime/rejected_logp": -1.2479492274433142,
+      "slime/rejected_stabilizer": 3.6173116623831447,
+      "slime/soft_gate": 0.49930259892717005,
+      "slime/stabilize_loss": 0.7234623459728027,
+      "source_mix/carefully": 0.1953125,
+      "source_mix/dolci-inst": 0.11796875,
+      "source_mix/dolci-think": 0.140625,
+      "source_mix/japanese": 0.11484375,
+      "source_mix/math": 0.11484375,
+      "source_mix/nvidia-cascade": 0.27109375,
+      "source_mix/self-intro": 0.0453125,
+      "step": 270
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 5.28125,
+      "learning_rate": 1.4e-06,
+      "loss": 1.171,
+      "slime/anchor_loss": 0.49193251332762883,
+      "slime/chosen_len": 1126.953125,
+      "slime/chosen_logp": -1.0931833913637092,
+      "slime/chosen_sft_loss": 1.0931833913637092,
+      "slime/chosen_sft_weighted_loss": 0.021863667332263505,
+      "slime/dist_loss": 0.01778693534189415,
+      "slime/hard_satisfied": 0.2859375,
+      "slime/loss": 1.1709950015967479,
+      "slime/margin": 0.10439605545252562,
+      "slime/rejected_len": 2512.12578125,
+      "slime/rejected_logp": -1.1975794465448417,
+      "slime/rejected_stabilizer": 3.197059365507448,
+      "slime/soft_gate": 0.4998471271479502,
+      "slime/stabilize_loss": 0.6394118847383652,
+      "source_mix/carefully": 0.1828125,
+      "source_mix/dolci-inst": 0.1234375,
+      "source_mix/dolci-think": 0.13671875,
+      "source_mix/japanese": 0.12890625,
+      "source_mix/math": 0.1078125,
+      "source_mix/nvidia-cascade": 0.28359375,
+      "source_mix/self-intro": 0.03671875,
+      "step": 280
+    },
+    {
+      "epoch": 0.9666666666666667,
+      "grad_norm": 5.5,
+      "learning_rate": 7.333333333333333e-07,
+      "loss": 1.2025,
+      "slime/anchor_loss": 0.4982497556033195,
+      "slime/chosen_len": 1129.396875,
+      "slime/chosen_logp": -1.107221708864381,
+      "slime/chosen_sft_loss": 1.107221708864381,
+      "slime/chosen_sft_weighted_loss": 0.02214443369007881,
+      "slime/dist_loss": 0.018779849886993816,
+      "slime/hard_satisfied": 0.27734375,
+      "slime/loss": 1.2024542617844418,
+      "slime/margin": 0.09719119273649994,
+      "slime/rejected_len": 2334.88671875,
+      "slime/rejected_logp": -1.2044129011283076,
+      "slime/rejected_stabilizer": 3.3164010618638713,
+      "slime/soft_gate": 0.5001242151716724,
+      "slime/stabilize_loss": 0.6632802235864801,
+      "source_mix/carefully": 0.190625,
+      "source_mix/dolci-inst": 0.121875,
+      "source_mix/dolci-think": 0.146875,
+      "source_mix/japanese": 0.13203125,
+      "source_mix/math": 0.10859375,
+      "source_mix/nvidia-cascade": 0.25625,
+      "source_mix/self-intro": 0.04375,
+      "step": 290
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 8.5,
+      "learning_rate": 6.666666666666665e-08,
+      "loss": 1.1436,
+      "slime/anchor_loss": 0.47786887729016597,
+      "slime/chosen_len": 1195.4171875,
+      "slime/chosen_logp": -1.0619308661553077,
+      "slime/chosen_sft_loss": 1.0619308661553077,
+      "slime/chosen_sft_weighted_loss": 0.02123861684704025,
+      "slime/dist_loss": 0.017122648745903746,
+      "slime/hard_satisfied": 0.2875,
+      "slime/loss": 1.1436487480183133,
+      "slime/margin": 0.12898020184657072,
+      "slime/rejected_len": 2849.75625,
+      "slime/rejected_logp": -1.19091106805281,
+      "slime/rejected_stabilizer": 3.1370929749740752,
+      "slime/soft_gate": 0.49892478859983386,
+      "slime/stabilize_loss": 0.6274186062939406,
+      "source_mix/carefully": 0.1734375,
+      "source_mix/dolci-inst": 0.1234375,
+      "source_mix/dolci-think": 0.140625,
+      "source_mix/japanese": 0.13203125,
+      "source_mix/math": 0.1359375,
+      "source_mix/nvidia-cascade": 0.2484375,
+      "source_mix/self-intro": 0.04609375,
+      "step": 300
+    },
+    {
+      "epoch": 1.0,
+      "step": 300,
+      "total_flos": 0.0,
+      "train_loss": 1.6715669790903727,
+      "train_runtime": 5088.5246,
+      "train_samples_per_second": 7.546,
+      "train_steps_per_second": 0.059
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 300,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfe37cc67b662ffee297b85bb7a30be0c4b498bb82bc9381e6a0dde81d5b4b84
+size 5969