Add tiny Anima modular pipeline fixture

README.md

@@ -0,0 +1,14 @@
+---
+library_name: diffusers
+tags:
+- modular-diffusers
+- anima
+---
+
+# tiny-anima-modular-pipe
+
+Tiny randomly-initialized Anima modular pipeline, used by the `diffusers` fast tests in
+`tests/modular_pipelines/anima/`. Not useful for generation.
+
+`tokenizer` and `t5_tokenizer` are not stored here — `modular_model_index.json` points them at
+`hf-internal-testing/tiny-random-Qwen2VLForConditionalGeneration` and `hf-internal-testing/tiny-random-t5`.

modular_model_index.json

@@ -0,0 +1,103 @@
+{
+  "_blocks_class_name": "AnimaAutoBlocks",
+  "_class_name": "AnimaModularPipeline",
+  "_diffusers_version": "0.40.0.dev0",
+  "scheduler": [
+    "diffusers",
+    "FlowMatchEulerDiscreteScheduler",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-anima-modular-pipe",
+      "revision": null,
+      "subfolder": "scheduler",
+      "type_hint": [
+        "diffusers",
+        "FlowMatchEulerDiscreteScheduler"
+      ],
+      "variant": null
+    }
+  ],
+  "t5_tokenizer": [
+    "transformers",
+    "T5TokenizerFast",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-random-t5",
+      "revision": null,
+      "subfolder": null,
+      "type_hint": [
+        "transformers",
+        "T5TokenizerFast"
+      ],
+      "variant": null
+    }
+  ],
+  "text_conditioner": [
+    "diffusers",
+    "AnimaTextConditioner",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-anima-modular-pipe",
+      "revision": null,
+      "subfolder": "text_conditioner",
+      "type_hint": [
+        "diffusers",
+        "AnimaTextConditioner"
+      ],
+      "variant": null
+    }
+  ],
+  "text_encoder": [
+    "transformers",
+    "Qwen3Model",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-anima-modular-pipe",
+      "revision": null,
+      "subfolder": "text_encoder",
+      "type_hint": [
+        "transformers",
+        "Qwen3Model"
+      ],
+      "variant": null
+    }
+  ],
+  "tokenizer": [
+    "transformers",
+    "Qwen2Tokenizer",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-random-Qwen2VLForConditionalGeneration",
+      "revision": null,
+      "subfolder": null,
+      "type_hint": [
+        "transformers",
+        "Qwen2Tokenizer"
+      ],
+      "variant": null
+    }
+  ],
+  "transformer": [
+    "diffusers",
+    "CosmosTransformer3DModel",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-anima-modular-pipe",
+      "revision": null,
+      "subfolder": "transformer",
+      "type_hint": [
+        "diffusers",
+        "CosmosTransformer3DModel"
+      ],
+      "variant": null
+    }
+  ],
+  "vae": [
+    "diffusers",
+    "AutoencoderKLQwenImage",
+    {
+      "pretrained_model_name_or_path": "hf-internal-testing/tiny-anima-modular-pipe",
+      "revision": null,
+      "subfolder": "vae",
+      "type_hint": [
+        "diffusers",
+        "AutoencoderKLQwenImage"
+      ],
+      "variant": null
+    }
+  ]
+}

scheduler/scheduler_config.json

@@ -0,0 +1,18 @@
+{
+  "_class_name": "FlowMatchEulerDiscreteScheduler",
+  "_diffusers_version": "0.40.0.dev0",
+  "base_image_seq_len": 256,
+  "base_shift": 0.5,
+  "invert_sigmas": false,
+  "max_image_seq_len": 4096,
+  "max_shift": 1.15,
+  "num_train_timesteps": 1000,
+  "shift": 3.0,
+  "shift_terminal": null,
+  "stochastic_sampling": false,
+  "time_shift_type": "exponential",
+  "use_beta_sigmas": false,
+  "use_dynamic_shifting": false,
+  "use_exponential_sigmas": false,
+  "use_karras_sigmas": false
+}

text_conditioner/config.json

@@ -0,0 +1,14 @@
+{
+  "_class_name": "AnimaTextConditioner",
+  "_diffusers_version": "0.40.0.dev0",
+  "min_sequence_length": 16,
+  "mlp_ratio": 4.0,
+  "model_dim": 16,
+  "num_attention_heads": 4,
+  "num_layers": 2,
+  "source_dim": 16,
+  "target_dim": 16,
+  "target_vocab_size": 32128,
+  "use_layer_norm": false,
+  "use_self_attention": true
+}

text_conditioner/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abd23e56f9608bb7065bcbfd5c5774db81746e47bc8aa34aeaa787a9b91e05a2
+size 2095080

text_encoder/config.json

@@ -0,0 +1,37 @@
+{
+  "architectures": [
+    "Qwen3Model"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "dtype": "float32",
+  "eos_token_id": null,
+  "head_dim": 4,
+  "hidden_act": "silu",
+  "hidden_size": 16,
+  "initializer_range": 0.02,
+  "intermediate_size": 32,
+  "layer_types": [
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 128,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 4,
+  "num_hidden_layers": 2,
+  "num_key_value_heads": 2,
+  "pad_token_id": null,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000.0,
+    "rope_type": "default"
+  },
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "5.14.1",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 152064
+}

text_encoder/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86192180357afb3f5a419f22b56461699b6a9138f53a88f33645d608657213af
+size 9753280

transformer/config.json

@@ -0,0 +1,36 @@
+{
+  "_class_name": "CosmosTransformer3DModel",
+  "_diffusers_version": "0.40.0.dev0",
+  "adaln_lora_dim": 4,
+  "attention_head_dim": 16,
+  "concat_padding_mask": true,
+  "controlnet_block_every_n": null,
+  "crossattn_proj_in_channels": 1024,
+  "encoder_hidden_states_channels": 1024,
+  "extra_pos_embed_type": null,
+  "img_context_dim_in": null,
+  "img_context_dim_out": 2048,
+  "img_context_num_tokens": 256,
+  "in_channels": 4,
+  "max_size": [
+    4,
+    32,
+    32
+  ],
+  "mlp_ratio": 2,
+  "num_attention_heads": 2,
+  "num_layers": 2,
+  "out_channels": 4,
+  "patch_size": [
+    1,
+    2,
+    2
+  ],
+  "rope_scale": [
+    1.0,
+    4.0,
+    4.0
+  ],
+  "text_embed_dim": 16,
+  "use_crossattn_projection": false
+}

transformer/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb50a8b9b263cb67ca09a7de8335a4236c0192434dc8f2bb751c4aadeb46f6fe
+size 130368

vae/config.json

@@ -0,0 +1,31 @@
+{
+  "_class_name": "AutoencoderKLQwenImage",
+  "_diffusers_version": "0.40.0.dev0",
+  "attn_scales": [],
+  "base_dim": 24,
+  "dim_mult": [
+    1,
+    2,
+    4
+  ],
+  "dropout": 0.0,
+  "input_channels": 3,
+  "latents_mean": [
+    0.0,
+    0.0,
+    0.0,
+    0.0
+  ],
+  "latents_std": [
+    1.0,
+    1.0,
+    1.0,
+    1.0
+  ],
+  "num_res_blocks": 1,
+  "temperal_downsample": [
+    false,
+    true
+  ],
+  "z_dim": 4
+}

vae/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d82f6ac6ca2e5767a17a5efc504e0470705b3b3318bab132dd218ff0ff342d29
+size 16230492