Add files using upload-large-folder tool

README.md

@@ -0,0 +1,159 @@
+---
+library_name: easydel
+pipeline_tag: image-to-text
+tags:
+  - easydel
+  - jax
+  - "qwen3_5"
+  - "ImageTextToText"
+  - "vanilla"
+---
+
+<p align="center">
+  <img alt="EasyDeL" src="https://raw.githubusercontent.com/erfanzar/easydel/main/images/easydel-logo-with-text.png" height="80">
+</p>
+
+<h1 align="center">Qwen/Qwen3.5-0.8B-Base</h1>
+
+<div align="center">
+  EasyDeL checkpoint converted from Qwen/Qwen3.5-0.8B-Base.
+</div>
+
+## Overview
+
+This checkpoint is intended to be loaded with EasyDeL on JAX (CPU/GPU/TPU). It supports sharded loading with `auto_shard_model=True` and configurable precision via `dtype`, `param_dtype`, and `precision`.
+
+## Quickstart
+
+```python
+import easydel as ed
+from jax import numpy as jnp, lax
+
+repo_id = "/dev/shm/conv/Qwen3.5-0.8B-Base"
+
+dtype = jnp.bfloat16  # try jnp.float16 on many GPUs
+
+model = ed.AutoEasyDeLModelForImageTextToText.from_pretrained(
+    repo_id,
+    dtype=dtype,
+    param_dtype=dtype,
+    precision=lax.Precision("fastest"),
+    sharding_axis_names=("dp", "fsdp", "ep", "tp", "sp"),
+    sharding_axis_dims=(1, -1, 1, 1, 1),
+    config_kwargs=ed.EasyDeLBaseConfigDict(
+        attn_dtype=dtype,
+        attn_mechanism=ed.AttentionMechanisms.VANILLA,
+        fsdp_is_ep_bound=True,
+        sp_is_ep_bound=True,
+        moe_method=ed.MoEMethods.FUSED_MOE,
+    ),
+    auto_shard_model=True,
+    partition_axis=ed.PartitionAxis(),
+)
+```
+
+If the repository only provides PyTorch weights, pass `from_torch=True` to `from_pretrained(...)`.
+
+## Sharding & Parallelism (Multi-Device)
+
+EasyDeL can scale to multiple devices by creating a logical device mesh. Most EasyDeL loaders use a 5D mesh:
+
+- `dp`: data parallel (replicated parameters, different batch shards)
+- `fsdp`: parameter sharding (memory saver; often the biggest axis)
+- `ep`: expert parallel (MoE; keep `1` for non-MoE models)
+- `tp`: tensor parallel (splits large matmuls)
+- `sp`: sequence parallel (splits sequence dimension)
+
+Use `sharding_axis_names=("dp","fsdp","ep","tp","sp")` and choose `sharding_axis_dims` so that their product equals your device count.
+You can use `-1` in `sharding_axis_dims` to let EasyDeL infer the remaining dimension.
+
+<details>
+<summary>Example sharding configs</summary>
+
+```python
+# 8 devices, pure FSDP
+sharding_axis_dims = (1, 8, 1, 1, 1)
+
+# 8 devices, 2-way DP x 4-way FSDP
+sharding_axis_dims = (2, 4, 1, 1, 1)
+
+# 8 devices, 4-way FSDP x 2-way TP
+sharding_axis_dims = (1, 4, 1, 2, 1)
+```
+</details>
+
+## Using via `eLargeModel` (ELM)
+
+`eLargeModel` is a higher-level interface that wires together loading, sharding, training, and eSurge inference from a single config.
+
+```python
+from easydel import eLargeModel
+
+repo_id = "/dev/shm/conv/Qwen3.5-0.8B-Base"
+
+elm = eLargeModel.from_pretrained(repo_id)  # task is auto-detected
+elm.set_dtype("bf16")
+elm.set_sharding(axis_names=("dp", "fsdp", "ep", "tp", "sp"), axis_dims=(1, -1, 1, 1, 1))
+
+model = elm.build_model()
+# Optional: build an inference engine
+# engine = elm.build_esurge()
+```
+
+<details>
+<summary>ELM YAML config example</summary>
+
+```yaml
+model:
+  name_or_path: "/dev/shm/conv/Qwen3.5-0.8B-Base"
+
+loader:
+  dtype: bf16
+  param_dtype: bf16
+
+sharding:
+  axis_dims: [1, -1, 1, 1, 1]
+  auto_shard_model: true
+```
+</details>
+
+## Features
+
+**EasyDeL:**
+- JAX native implementation and sharded execution
+- Configurable attention backends via `AttentionMechanisms.*`
+- Precision control via `dtype`, `param_dtype`, and `precision`
+
+## Installation
+
+```bash
+pip install easydel
+```
+
+## Links
+
+- EasyDeL GitHub: https://github.com/erfanzar/EasyDeL
+- Docs: https://easydel.readthedocs.io/en/latest/
+
+## Supported Tasks
+
+- ImageTextToText
+
+## Limitations
+
+- Refer to the original model card for training data, evaluation, and intended use.
+
+## License
+
+EasyDeL is released under the Apache-2.0 license. The license for this model's weights may differ; please consult the original repository.
+
+## Citation
+
+```bibtex
+@misc{Zare Chavoshi_2023,
+    title={EasyDeL: An open-source library for enhancing and streamlining the training process of machine learning models},
+    url={https://github.com/erfanzar/EasyDeL},
+    author={Zare Chavoshi, Erfan},
+    year={2023}
+}
+```

chat_template.jinja

@@ -0,0 +1,154 @@
+{%- set image_count = namespace(value=0) %}
+{%- set video_count = namespace(value=0) %}
+{%- macro render_content(content, do_vision_count, is_system_content=false) %}
+    {%- if content is string %}
+        {{- content }}
+    {%- elif content is iterable and content is not mapping %}
+        {%- for item in content %}
+            {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain images.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set image_count.value = image_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Picture ' ~ image_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
+            {%- elif 'video' in item or item.type == 'video' %}
+                {%- if is_system_content %}
+                    {{- raise_exception('System message cannot contain videos.') }}
+                {%- endif %}
+                {%- if do_vision_count %}
+                    {%- set video_count.value = video_count.value + 1 %}
+                {%- endif %}
+                {%- if add_vision_id %}
+                    {{- 'Video ' ~ video_count.value ~ ': ' }}
+                {%- endif %}
+                {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
+            {%- elif 'text' in item %}
+                {{- item.text }}
+            {%- else %}
+                {{- raise_exception('Unexpected item type in content.') }}
+            {%- endif %}
+        {%- endfor %}
+    {%- elif content is none or content is undefined %}
+        {{- '' }}
+    {%- else %}
+        {{- raise_exception('Unexpected content type.') }}
+    {%- endif %}
+{%- endmacro %}
+{%- if not messages %}
+    {{- raise_exception('No messages provided.') }}
+{%- endif %}
+{%- if tools and tools is iterable and tools is not mapping %}
+    {{- '<|im_start|>system\n' }}
+    {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>" }}
+    {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {%- if content %}
+            {{- '\n\n' + content }}
+        {%- endif %}
+    {%- endif %}
+    {{- '<|im_end|>\n' }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {%- set content = render_content(messages[0].content, false, true)|trim %}
+        {{- '<|im_start|>system\n' + 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" %}
+        {%- set content = render_content(message.content, false)|trim %}
+        {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
+            {%- set ns.multi_step_tool = false %}
+            {%- set ns.last_query_index = index %}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if ns.multi_step_tool %}
+    {{- raise_exception('No user query found in messages.') }}
+{%- endif %}
+{%- for message in messages %}
+    {%- set content = render_content(message.content, true)|trim %}
+    {%- if message.role == "system" %}
+        {%- if not loop.first %}
+            {{- raise_exception('System message must be at the beginning.') }}
+        {%- endif %}
+    {%- elif message.role == "user" %}
+        {{- '<|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 %}
+        {%- set reasoning_content = reasoning_content|trim %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if tool_call.function is defined %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {%- if loop.first %}
+                    {%- if content|trim %}
+                        {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- else %}
+                        {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                    {%- endif %}
+                {%- else %}
+                    {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
+                {%- endif %}
+                {%- if tool_call.arguments is defined %}
+                    {%- for args_name, args_value in tool_call.arguments|items %}
+                        {{- '<parameter=' + args_name + '>\n' }}
+                        {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
+                        {{- args_value }}
+                        {{- '\n</parameter>\n' }}
+                    {%- endfor %}
+                {%- endif %}
+                {{- '</function>\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.previtem and loop.previtem.role != "tool" %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if not loop.last and loop.nextitem.role != "tool" %}
+            {{- '<|im_end|>\n' }}
+        {%- elif loop.last %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- else %}
+        {{- raise_exception('Unexpected message role.') }}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is true %}
+        {{- '<think>\n' }}
+    {%- else %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}

config.json

@@ -0,0 +1,473 @@
+{
+  "_external_rope_config_kwargs": {},
+  "add_cross_attention": false,
+  "architectures": [
+    "Qwen3_5ForConditionalGeneration"
+  ],
+  "attn_mechanism": "vanilla",
+  "backend": null,
+  "bits": null,
+  "blocksize_b": 1,
+  "blocksize_k": 512,
+  "blocksize_q": 512,
+  "bos_token_id": null,
+  "cross_attention_hidden_size": null,
+  "decode_attn_mechanism": null,
+  "decoder_start_token_id": null,
+  "easy_method": "train",
+  "eos_token_id": null,
+  "fcm_max_ratio": 0.0,
+  "fcm_min_ratio": 0.0,
+  "flash_attention_backward_pass_impl": "triton",
+  "fsdp_is_ep_bound": true,
+  "gradient_checkpointing": "",
+  "gradient_checkpointing_targets": null,
+  "hardware_abstraction": false,
+  "image_token_id": 248056,
+  "is_decoder": false,
+  "kv_cache_quantization_config": null,
+  "kv_cache_sharding_sequence_axis_name": "sp",
+  "kvdtype": "bfloat16",
+  "lmhead_chunksize": null,
+  "max_position_embeddings": null,
+  "mla_attn_dtype": "bfloat16",
+  "mla_attn_mechanism": "auto",
+  "mla_attn_softmax_dtype": "float32",
+  "model_type": "qwen3_5",
+  "moe_force_xla_gmm": false,
+  "moe_method": "fused_moe",
+  "moe_tiling_size_batch": 4,
+  "moe_tiling_size_dim": 128,
+  "moe_tiling_size_seqlen": 128,
+  "operation_configs": null,
+  "pad_token_id": null,
+  "pallas_k_block_size": 128,
+  "pallas_m_block_size": 128,
+  "pallas_n_block_size": 128,
+  "partition_axis": {
+    "attention_dim_axis": null,
+    "attention_kv_dim_axis": null,
+    "batch_axis": [
+      "fsdp",
+      "dp"
+    ],
+    "bias_head_sequence_axis": null,
+    "bias_key_sequence_axis": null,
+    "data_parallel_axis": "dp",
+    "decode_attention_dim_axis": null,
+    "decode_attention_kv_dim_axis": null,
+    "decode_batch_axis": [
+      "fsdp",
+      "dp"
+    ],
+    "decode_head_axis": "tp",
+    "decode_key_sequence_axis": "sp",
+    "decode_kv_head_axis": "tp",
+    "decode_query_sequence_axis": null,
+    "expert_axis": "ep",
+    "expert_gate_axis": null,
+    "expert_parallel_axis": "ep",
+    "fully_sharded_data_parallel_axis": "fsdp",
+    "head_axis": "tp",
+    "hidden_state_axis": "tp",
+    "key_sequence_axis": "sp",
+    "kv_head_axis": "tp",
+    "mlp_intermediate_axis": "tp",
+    "query_sequence_axis": "sp",
+    "sequence_axis": "sp",
+    "sequence_parallel_axis": "sp",
+    "tensor_parallel_axis": "tp",
+    "vocab_axis": "tp"
+  },
+  "platform": null,
+  "precompute_masks": true,
+  "pretraining_tp": 1,
+  "qmm_platform_override": null,
+  "qmm_tpu_path_override": null,
+  "quantization_config": null,
+  "scan_attention_layers": false,
+  "scan_mlp_chunk_size": 1024,
+  "scan_ring_attention": true,
+  "sep_token_id": null,
+  "sequence_axis_name": "sp",
+  "sharding_axis_dims": [
+    1,
+    -1,
+    1,
+    1,
+    1
+  ],
+  "sharding_axis_names": [
+    "dp",
+    "fsdp",
+    "ep",
+    "tp",
+    "sp"
+  ],
+  "sharding_dcn_axis_dims": null,
+  "sp_is_ep_bound": true,
+  "text_config": {
+    "_external_rope_config_kwargs": {},
+    "add_cross_attention": false,
+    "architectures": [
+      "Qwen3_5ForConditionalGeneration"
+    ],
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "attn_dtype": "bfloat16",
+    "attn_mechanism": "vanilla",
+    "attn_output_gate": true,
+    "attn_softmax_dtype": "float32",
+    "backend": null,
+    "bits": null,
+    "blocksize_b": 1,
+    "blocksize_k": 512,
+    "blocksize_q": 512,
+    "bos_token_id": null,
+    "cross_attention_hidden_size": null,
+    "decode_attn_mechanism": null,
+    "decoder_sparse_step": 1,
+    "decoder_start_token_id": null,
+    "dtype": "bfloat16",
+    "easy_method": "train",
+    "eos_token_id": 248044,
+    "fcm_max_ratio": 0.0,
+    "fcm_min_ratio": 0.0,
+    "flash_attention_backward_pass_impl": "triton",
+    "fsdp_is_ep_bound": true,
+    "full_attention_interval": 4,
+    "gradient_checkpointing": "",
+    "gradient_checkpointing_targets": null,
+    "hardware_abstraction": false,
+    "head_dim": 256,
+    "hidden_act": "silu",
+    "hidden_size": 1024,
+    "initializer_range": 0.02,
+    "intermediate_size": 3584,
+    "is_decoder": false,
+    "kv_cache_quantization_config": null,
+    "kv_cache_sharding_sequence_axis_name": "sp",
+    "kvdtype": "bfloat16",
+    "layer_types": [
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention",
+      "linear_attention",
+      "linear_attention",
+      "linear_attention",
+      "full_attention"
+    ],
+    "linear_attention_separate_proj": true,
+    "linear_conv_kernel_dim": 4,
+    "linear_key_head_dim": 128,
+    "linear_num_key_heads": 16,
+    "linear_num_value_heads": 16,
+    "linear_value_head_dim": 128,
+    "lmhead_chunksize": null,
+    "mamba_ssm_dtype": "float32",
+    "max_position_embeddings": 262144,
+    "mla_attn_dtype": "bfloat16",
+    "mla_attn_mechanism": "auto",
+    "mla_attn_softmax_dtype": "float32",
+    "mlp_only_layers": [
+      0,
+      1,
+      2,
+      3,
+      4,
+      5,
+      6,
+      7,
+      8,
+      9,
+      10,
+      11,
+      12,
+      13,
+      14,
+      15,
+      16,
+      17,
+      18,
+      19,
+      20,
+      21,
+      22,
+      23
+    ],
+    "model_type": "qwen3_5_text",
+    "moe_force_xla_gmm": false,
+    "moe_intermediate_size": 512,
+    "moe_method": "fused_moe",
+    "moe_tiling_size_batch": 4,
+    "moe_tiling_size_dim": 128,
+    "moe_tiling_size_seqlen": 128,
+    "mtp_num_hidden_layers": 1,
+    "mtp_use_dedicated_embeddings": false,
+    "norm_topk_prob": true,
+    "num_attention_heads": 8,
+    "num_experts": 256,
+    "num_experts_per_tok": 8,
+    "num_hidden_layers": 24,
+    "num_key_value_heads": 2,
+    "num_local_experts": 256,
+    "operation_configs": null,
+    "output_router_logits": false,
+    "pad_token_id": null,
+    "pallas_k_block_size": 128,
+    "pallas_m_block_size": 128,
+    "pallas_n_block_size": 128,
+    "partial_rotary_factor": 0.25,
+    "partition_axis": {
+      "attention_dim_axis": null,
+      "attention_kv_dim_axis": null,
+      "batch_axis": [
+        "fsdp",
+        "dp"
+      ],
+      "bias_head_sequence_axis": null,
+      "bias_key_sequence_axis": null,
+      "data_parallel_axis": "dp",
+      "decode_attention_dim_axis": null,
+      "decode_attention_kv_dim_axis": null,
+      "decode_batch_axis": [
+        "fsdp",
+        "dp"
+      ],
+      "decode_head_axis": "tp",
+      "decode_key_sequence_axis": "sp",
+      "decode_kv_head_axis": "tp",
+      "decode_query_sequence_axis": null,
+      "expert_axis": "ep",
+      "expert_gate_axis": null,
+      "expert_parallel_axis": "ep",
+      "fully_sharded_data_parallel_axis": "fsdp",
+      "head_axis": "tp",
+      "hidden_state_axis": "tp",
+      "key_sequence_axis": "sp",
+      "kv_head_axis": "tp",
+      "mlp_intermediate_axis": "tp",
+      "query_sequence_axis": "sp",
+      "sequence_axis": "sp",
+      "sequence_parallel_axis": "sp",
+      "tensor_parallel_axis": "tp",
+      "vocab_axis": "tp"
+    },
+    "platform": null,
+    "precompute_masks": true,
+    "pretraining_tp": 1,
+    "qmm_platform_override": null,
+    "qmm_tpu_path_override": null,
+    "quantization_config": null,
+    "rms_norm_eps": 1e-06,
+    "rope_parameters": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        11,
+        11,
+        10
+      ],
+      "partial_rotary_factor": 0.25,
+      "rope_theta": 10000000,
+      "rope_type": "default",
+      "type": "default"
+    },
+    "rope_theta": 10000000,
+    "router_aux_loss_coef": 0.001,
+    "scan_attention_layers": false,
+    "scan_mlp_chunk_size": 1024,
+    "scan_ring_attention": true,
+    "sep_token_id": null,
+    "sequence_axis_name": "sp",
+    "sharding_axis_dims": [
+      1,
+      -1,
+      1,
+      1,
+      1
+    ],
+    "sharding_axis_names": [
+      "dp",
+      "fsdp",
+      "ep",
+      "tp",
+      "sp"
+    ],
+    "sharding_dcn_axis_dims": null,
+    "shared_expert_intermediate_size": 512,
+    "sp_is_ep_bound": true,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "use_cache": true,
+    "use_expert_tensor_mode": false,
+    "use_qmm_best_config": false,
+    "use_ring_of_experts": false,
+    "use_scan_mlp": false,
+    "use_sharded_kv_caching": false,
+    "use_sharding_constraint": false,
+    "vocab_size": 248320
+  },
+  "tie_encoder_decoder": false,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.5.0",
+  "use_expert_tensor_mode": false,
+  "use_qmm_best_config": false,
+  "use_ring_of_experts": false,
+  "use_scan_mlp": false,
+  "use_sharded_kv_caching": false,
+  "use_sharding_constraint": false,
+  "video_token_id": 248057,
+  "vision_config": {
+    "_external_rope_config_kwargs": {},
+    "add_cross_attention": false,
+    "architectures": [
+      "Qwen3_5ForConditionalGeneration"
+    ],
+    "attn_dtype": "bfloat16",
+    "attn_mechanism": "vanilla",
+    "attn_softmax_dtype": "float32",
+    "backend": null,
+    "bits": null,
+    "blocksize_b": 1,
+    "blocksize_k": 512,
+    "blocksize_q": 512,
+    "bos_token_id": null,
+    "cross_attention_hidden_size": null,
+    "decode_attn_mechanism": null,
+    "decoder_start_token_id": null,
+    "deepstack_visual_indexes": [],
+    "depth": 12,
+    "easy_method": "train",
+    "embed_dim": 768,
+    "eos_token_id": null,
+    "fcm_max_ratio": 0.0,
+    "fcm_min_ratio": 0.0,
+    "flash_attention_backward_pass_impl": "triton",
+    "fsdp_is_ep_bound": true,
+    "gradient_checkpointing": "",
+    "gradient_checkpointing_targets": null,
+    "hardware_abstraction": false,
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 768,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 3072,
+    "is_decoder": false,
+    "kv_cache_quantization_config": null,
+    "kv_cache_sharding_sequence_axis_name": "sp",
+    "kvdtype": "bfloat16",
+    "lmhead_chunksize": null,
+    "max_position_embeddings": null,
+    "mla_attn_dtype": "bfloat16",
+    "mla_attn_mechanism": "auto",
+    "mla_attn_softmax_dtype": "float32",
+    "model_type": "qwen3_5",
+    "moe_force_xla_gmm": false,
+    "moe_method": "fused_moe",
+    "moe_tiling_size_batch": 4,
+    "moe_tiling_size_dim": 128,
+    "moe_tiling_size_seqlen": 128,
+    "num_attention_heads": 12,
+    "num_heads": 12,
+    "num_position_embeddings": 2304,
+    "operation_configs": null,
+    "out_hidden_size": 1024,
+    "pad_token_id": null,
+    "pallas_k_block_size": 128,
+    "pallas_m_block_size": 128,
+    "pallas_n_block_size": 128,
+    "partition_axis": {
+      "attention_dim_axis": null,
+      "attention_kv_dim_axis": null,
+      "batch_axis": [
+        "fsdp",
+        "dp"
+      ],
+      "bias_head_sequence_axis": null,
+      "bias_key_sequence_axis": null,
+      "data_parallel_axis": "dp",
+      "decode_attention_dim_axis": null,
+      "decode_attention_kv_dim_axis": null,
+      "decode_batch_axis": [
+        "fsdp",
+        "dp"
+      ],
+      "decode_head_axis": "tp",
+      "decode_key_sequence_axis": "sp",
+      "decode_kv_head_axis": "tp",
+      "decode_query_sequence_axis": null,
+      "expert_axis": "ep",
+      "expert_gate_axis": null,
+      "expert_parallel_axis": "ep",
+      "fully_sharded_data_parallel_axis": "fsdp",
+      "head_axis": "tp",
+      "hidden_state_axis": "tp",
+      "key_sequence_axis": "sp",
+      "kv_head_axis": "tp",
+      "mlp_intermediate_axis": "tp",
+      "query_sequence_axis": "sp",
+      "sequence_axis": "sp",
+      "sequence_parallel_axis": "sp",
+      "tensor_parallel_axis": "tp",
+      "vocab_axis": "tp"
+    },
+    "patch_size": 16,
+    "platform": null,
+    "precompute_masks": true,
+    "pretraining_tp": 1,
+    "qmm_platform_override": null,
+    "qmm_tpu_path_override": null,
+    "quantization_config": null,
+    "scan_attention_layers": false,
+    "scan_mlp_chunk_size": 1024,
+    "scan_ring_attention": true,
+    "sep_token_id": null,
+    "sequence_axis_name": "sp",
+    "sharding_axis_dims": [
+      1,
+      -1,
+      1,
+      1,
+      1
+    ],
+    "sharding_axis_names": [
+      "dp",
+      "fsdp",
+      "ep",
+      "tp",
+      "sp"
+    ],
+    "sharding_dcn_axis_dims": null,
+    "sp_is_ep_bound": true,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "tokens_per_second": 2.0,
+    "use_expert_tensor_mode": false,
+    "use_qmm_best_config": false,
+    "use_ring_of_experts": false,
+    "use_scan_mlp": false,
+    "use_sharded_kv_caching": false,
+    "use_sharding_constraint": false
+  },
+  "vision_end_token_id": 248054,
+  "vision_start_token_id": 248053
+}

model/model/language_model/embed_tokens/embedding/.zarray

@@ -0,0 +1 @@
+{"chunks":[62080,1024],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[248320,1024],"zarr_format":2}

model/model/language_model/layers/0/input_layernorm/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[1024],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024],"zarr_format":2}

model/model/language_model/layers/0/linear_attn/A_log/.zarray

@@ -0,0 +1 @@
+{"chunks":[16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[16],"zarr_format":2}

model/model/language_model/layers/0/linear_attn/conv1d/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,1,6144],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[4,1,6144],"zarr_format":2}

model/model/language_model/layers/0/linear_attn/dt_bias/.zarray

@@ -0,0 +1 @@
+{"chunks":[16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[16],"zarr_format":2}

model/model/language_model/layers/0/linear_attn/in_proj_a/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,16],"zarr_format":2}

model/model/language_model/layers/19/post_attention_layernorm/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[1024],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024],"zarr_format":2}

model/model/language_model/layers/19/self_attn/k_proj/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,512],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,512],"zarr_format":2}

model/model/language_model/layers/19/self_attn/q_norm/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[256],"zarr_format":2}

model/model/language_model/layers/19/self_attn/v_proj/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,512],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,512],"zarr_format":2}

model/model/language_model/layers/2/input_layernorm/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[1024],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024],"zarr_format":2}

model/model/language_model/layers/2/linear_attn/conv1d/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,1,6144],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[4,1,6144],"zarr_format":2}

model/model/language_model/layers/2/linear_attn/in_proj_b/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,16],"zarr_format":2}

model/model/language_model/layers/2/linear_attn/in_proj_z/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,2048],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,2048],"zarr_format":2}

model/model/language_model/layers/2/mlp/down_proj/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[3584,256],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[3584,1024],"zarr_format":2}

model/model/language_model/layers/2/mlp/up_proj/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,3584],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,3584],"zarr_format":2}

model/model/language_model/layers/2/post_attention_layernorm/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[1024],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024],"zarr_format":2}

model/model/language_model/layers/20/input_layernorm/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[1024],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024],"zarr_format":2}

model/model/language_model/layers/20/linear_attn/A_log/.zarray

@@ -0,0 +1 @@
+{"chunks":[16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[16],"zarr_format":2}

model/model/language_model/layers/20/linear_attn/conv1d/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,1,6144],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[4,1,6144],"zarr_format":2}

model/model/language_model/layers/20/linear_attn/dt_bias/.zarray

@@ -0,0 +1 @@
+{"chunks":[16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[16],"zarr_format":2}

model/model/language_model/layers/20/linear_attn/in_proj_a/kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,16],"compressor":{"id":"zstd","level":1},"dimension_separator":".","dtype":"bfloat16","fill_value":null,"filters":null,"order":"C","shape":[1024,16],"zarr_format":2}