Upload folder using huggingface_hub

lam/config.json

@@ -0,0 +1,78 @@
+{
+  "action_depth": 5,
+  "action_dropout": 0.0,
+  "action_hidden_dim": 96,
+  "action_obs_dim": 0,
+  "action_prev_dim": 10,
+  "action_state_dim": 5,
+  "action_target_dim": 10,
+  "action_wide_dim": 512,
+  "architectures": [
+    "LAMModel"
+  ],
+  "decoder_attention_head_dim": 64,
+  "decoder_attn_implementation": "flash_attention_2",
+  "decoder_encoder_hidden_dim": 6,
+  "decoder_eps": 1e-06,
+  "decoder_ffn_dim": 768,
+  "decoder_freq_dim": 64,
+  "decoder_in_channels": 3,
+  "decoder_num_attention_heads": 3,
+  "decoder_num_layers": 12,
+  "decoder_out_channels": 3,
+  "decoder_patch_size": [
+    4,
+    4
+  ],
+  "decoder_pos_embed_seq_len": null,
+  "decoder_rope_max_seq_len": 1024,
+  "dtype": "bfloat16",
+  "encoder_height": 64,
+  "encoder_width": 64,
+  "fsq_levels": [
+    8,
+    8,
+    8,
+    5,
+    5,
+    5
+  ],
+  "initializer_range": 0.02,
+  "is_diffusion": true,
+  "latent_channels": 6,
+  "max_tokens": 256,
+  "min_tokens": 1,
+  "model_type": "lam",
+  "null_latent": 0,
+  "transformers_version": "4.57.1",
+  "use_tail_drop": true,
+  "videomae_config": {
+    "attn_drop_rate": 0.0,
+    "cos_attn": false,
+    "depth": 8,
+    "drop_path_rate": 0.0,
+    "drop_rate": 0.0,
+    "embed_dim": 192,
+    "img_size": [
+      64,
+      64
+    ],
+    "in_chans": 3,
+    "init_values": 0.0,
+    "layer_norm_eps": 1e-06,
+    "mlp_ratio": 4,
+    "norm_layer": "nn.LayerNorm",
+    "num_classes": 0,
+    "num_frames": 2,
+    "num_heads": 3,
+    "patch_size": 4,
+    "qk_scale": null,
+    "qkv_bias": true,
+    "tubelet_size": 2,
+    "use_learnable_pos_emb": false,
+    "use_mean_pooling": false,
+    "with_cp": false
+  },
+  "videomae_from_pretrained": null,
+  "vocab_size": 64000
+}

lam/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:946f7eeff919289fca9a0a6b37d1bd3787f4268c8b1146fa7f61d1ea137cddc0
+size 24600668

model_index.json

@@ -0,0 +1,24 @@
+{
+  "_class_name": "LAMPolicyPipeline",
+  "_diffusers_version": "0.35.2",
+  "lam": [
+    "flexlam_mini.models.lam.modeling_lam",
+    "LAMModel"
+  ],
+  "policy": [
+    "flexlam_mini.models.policy.modeling_policy",
+    "PolicyQwen3ForConditionalGeneration"
+  ],
+  "policy_processor": [
+    "transformers",
+    "Qwen2VLImageProcessor"
+  ],
+  "processor": [
+    "flexlam_mini.models.lam.processing_lam",
+    "LAMProcessorFast"
+  ],
+  "scheduler": [
+    "diffusers",
+    "FlowMatchEulerDiscreteScheduler"
+  ]
+}

policy/config.json

@@ -0,0 +1,51 @@
+{
+  "action_latent_dim": 6,
+  "action_seq_len": 256,
+  "action_start_token_id": 64003,
+  "action_vocab_size": 64000,
+  "architectures": [
+    "PolicyQwen3ForConditionalGeneration"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "dtype": "bfloat16",
+  "eos_token_id": 64004,
+  "head_dim": 64,
+  "hidden_act": "silu",
+  "hidden_size": 192,
+  "image_token_id": 64002,
+  "initializer_range": 0.02,
+  "intermediate_size": 768,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 2048,
+  "max_window_layers": 28,
+  "model_type": "policy_qwen3",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 8,
+  "num_key_value_heads": 3,
+  "pad_token_id": 0,
+  "predict_tokens": false,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 10000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": false,
+  "transformers_version": "4.57.1",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vision_end_token_id": 64001,
+  "vision_in_channels": 3,
+  "vision_merge_size": 2,
+  "vision_patch_size": 4,
+  "vision_start_token_id": 64000,
+  "vocab_size": 64005
+}

policy/generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "eos_token_id": 64004,
+  "pad_token_id": 0,
+  "transformers_version": "4.57.1",
+  "use_cache": false
+}

policy/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17f10d11358eec5b52f26e047e42bb792e02b9e453fdfd9460f0d01f876ad724
+size 58632988

policy_processor/preprocessor_config.json

@@ -0,0 +1,28 @@
+{
+  "do_convert_rgb": true,
+  "do_normalize": true,
+  "do_rescale": true,
+  "do_resize": true,
+  "image_mean": [
+    0.48145466,
+    0.4578275,
+    0.40821073
+  ],
+  "image_processor_type": "Qwen2VLImageProcessor",
+  "image_std": [
+    0.26862954,
+    0.26130258,
+    0.27577711
+  ],
+  "max_pixels": 1003520,
+  "merge_size": 2,
+  "min_pixels": 3136,
+  "patch_size": 4,
+  "resample": 3,
+  "rescale_factor": 0.00392156862745098,
+  "size": {
+    "longest_edge": 1003520,
+    "shortest_edge": 3136
+  },
+  "temporal_patch_size": 1
+}

processor/processor_config.json

@@ -0,0 +1,7 @@
+{
+  "encoder_height": 64,
+  "encoder_width": 64,
+  "height": 64,
+  "processor_class": "LAMProcessorFast",
+  "width": 64
+}

scheduler/scheduler_config.json

@@ -0,0 +1,18 @@
+{
+  "_class_name": "FlowMatchEulerDiscreteScheduler",
+  "_diffusers_version": "0.35.2",
+  "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": 1.0,
+  "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
+}