Text Generation
Transformers
Safetensors
metis16
metis
post-trained
instruction-tuned
reasoning
mixture-of-experts
mamba
conversational
custom_code
Instructions to use Lernex/Metis-1.6-Logos with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Lernex/Metis-1.6-Logos with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="Lernex/Metis-1.6-Logos", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("Lernex/Metis-1.6-Logos", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Lernex/Metis-1.6-Logos with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Lernex/Metis-1.6-Logos" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Lernex/Metis-1.6-Logos", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/Lernex/Metis-1.6-Logos
- SGLang
How to use Lernex/Metis-1.6-Logos with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "Lernex/Metis-1.6-Logos" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Lernex/Metis-1.6-Logos", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "Lernex/Metis-1.6-Logos" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Lernex/Metis-1.6-Logos", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use Lernex/Metis-1.6-Logos with Docker Model Runner:
docker model run hf.co/Lernex/Metis-1.6-Logos
Download eval_native_config.json from Lernex/Metis-1.6-Logos: direct link, hf CLI and curl.
- Browser
- Download file 5.94 kB
-
https://huggingface.co/Lernex/Metis-1.6-Logos/resolve/main/eval_native_config.json
- Command line
-
hf download hf://Lernex/Metis-1.6-Logos/eval_native_config.json
-
curl -L -o eval_native_config.json https://huggingface.co/Lernex/Metis-1.6-Logos/resolve/main/eval_native_config.json
5.94 kB
| { | |
| "activation_recompute_policy": "layer", | |
| "attention_backend": "varlen_fused_required", | |
| "attention_indices": [ | |
| 5, | |
| 10, | |
| 15 | |
| ], | |
| "attention_pass_lora_rank": 8, | |
| "autotune": { | |
| "compile_modes": [ | |
| "none" | |
| ], | |
| "dispatch_overlap": [ | |
| "on" | |
| ], | |
| "gates": { | |
| "max_fp8_loss_relative_error": 0.02, | |
| "max_grad_norm": 100.0, | |
| "max_hbm_fraction": 0.9, | |
| "max_ngram_layout_loss_relative_error": 0.001, | |
| "max_update_to_weight_ratio": 0.01 | |
| }, | |
| "global_token_batch": { | |
| "maximum": 12582912, | |
| "minimum": 1572864, | |
| "target": 6291456 | |
| }, | |
| "grad_accum_steps": [ | |
| 1 | |
| ], | |
| "learning_rates": [ | |
| 0.00012 | |
| ], | |
| "micro_batch_sizes": [ | |
| 4, | |
| 2 | |
| ], | |
| "ngram_table_modes": [ | |
| "row_sharded" | |
| ], | |
| "precision_profiles": [ | |
| "bf16" | |
| ], | |
| "preferred_learning_rate": 0.00012, | |
| "status": "bounded_canary_candidates_not_validated_optima" | |
| }, | |
| "bucket_varlen_attention": false, | |
| "budget_controller_leak": 0.0, | |
| "budget_controller_rate": 0.0, | |
| "budget_multiplier_limit": 1000.0, | |
| "budgeted_depth_calibration_coefficient": 0.0, | |
| "budgeted_depth_values": [], | |
| "budgeted_k_calibration_coefficient": 0.0, | |
| "causal_reference_ceiling": [], | |
| "context_extension_train_length": 163840, | |
| "context_parallel_size": 1, | |
| "continuation_entropy_coefficient": 0.001, | |
| "continuation_gate_init": 0.0, | |
| "d_model": 2560, | |
| "dense_ffn_intermediate_dim": 0, | |
| "depth_budget_coefficient": 0.01, | |
| "depth_budget_controller_rate": 0.0, | |
| "expected_parameter_audit": { | |
| "active_per_pass_max": 1378197900, | |
| "active_per_pass_mean": 1189454220, | |
| "active_per_pass_min": 1047896460, | |
| "continuation": 2032129, | |
| "depth_memory": 4211972, | |
| "embedding": 167772160, | |
| "expert_routers": 6107040, | |
| "final_norm": 2560, | |
| "latent_projections": 104857600, | |
| "mhc": 176320, | |
| "mixers": 833409535, | |
| "ngram_fusion": 2726440, | |
| "ngram_tables": 1771823360, | |
| "routed_experts": 9059696640, | |
| "shared_experts": 47185920, | |
| "stored_total": 12000001676 | |
| }, | |
| "expert_balance_bias_update_rate": 0.001, | |
| "expert_balance_coefficient": 0.0, | |
| "expert_execution": "grouped_gemm", | |
| "expert_gate_normalisation": "selected", | |
| "expert_intermediate_dim": 768, | |
| "expert_parallel_size": 1, | |
| "expert_replicas": 1, | |
| "expert_swiglu_fused": false, | |
| "expert_weight_chunks": 1, | |
| "family": "logos", | |
| "ffn_mode": "moe", | |
| "final_context_length": 131072, | |
| "head_dim": 64, | |
| "k_budget_coefficient": 0.01, | |
| "latent_dim": 1024, | |
| "lm_head_chunk_size": 1024, | |
| "mamba_backend": "fused_required", | |
| "mamba_chunk_size": 256, | |
| "mamba_d_conv": 4, | |
| "mamba_d_state": 128, | |
| "mamba_expand": 2, | |
| "mamba_head_dim": 64, | |
| "mamba_ngroups": 10, | |
| "max_passes": 5, | |
| "max_routed_k": 8, | |
| "memory_dim": 256, | |
| "memory_gate_init": -6.0, | |
| "memory_heads": 4, | |
| "mhc_backend": "fused_required", | |
| "mhc_pass_embedding_dim": 64, | |
| "mhc_sinkhorn_iterations": 8, | |
| "min_routed_k": 1, | |
| "model_type": "metis16_more", | |
| "moe_dispatch_capacity_factor": 0.0, | |
| "moe_dispatch_chunks": 1, | |
| "n_heads": 40, | |
| "n_kv_heads": 10, | |
| "n_layers": 20, | |
| "n_routed_experts": 192, | |
| "n_shared_experts": 1, | |
| "n_streams": 4, | |
| "name": "Metis-1.6-Logos", | |
| "ngram_memory": { | |
| "canonicalization": "tokenizer_sidecar_required", | |
| "hash_seeds": [ | |
| 2654435761, | |
| 2246822519, | |
| 3266489917, | |
| 668265263, | |
| 374761393, | |
| 3550635117, | |
| 4251991749, | |
| 3043081225 | |
| ], | |
| "injection_layers": [ | |
| 2, | |
| 5 | |
| ], | |
| "learning_rate_scale": 2.0, | |
| "optimizer": "sparse_adam", | |
| "orders": [ | |
| 2, | |
| 3 | |
| ], | |
| "slots_by_order": { | |
| "2": [ | |
| 1723303, | |
| 1724299, | |
| 1725301, | |
| 1726303, | |
| 1727317, | |
| 1728317, | |
| 1729307, | |
| 1730101 | |
| ], | |
| "3": [ | |
| 1730507, | |
| 1731311, | |
| 1732301, | |
| 1733297, | |
| 1734311, | |
| 1735301, | |
| 1736303, | |
| 1737161 | |
| ] | |
| }, | |
| "sparse_gradients": false, | |
| "table_mode": "replicated", | |
| "tables_per_order": 8, | |
| "value_dim": 64, | |
| "weight_decay": 0.0 | |
| }, | |
| "no_position_embeddings": true, | |
| "precision": { | |
| "activation_dtype": "bfloat16", | |
| "allow_bf16_fallback": true, | |
| "backend": "auto", | |
| "bf16_roles": [ | |
| "embedding", | |
| "ngram_table", | |
| "residual_stream", | |
| "mamba_state", | |
| "normalization", | |
| "gate", | |
| "collective", | |
| "attention_pass_lora_down", | |
| "attention_pass_lora_up", | |
| "memory_metadata_write_projection" | |
| ], | |
| "expert_collective_wire": "fp8", | |
| "fp32_roles": [ | |
| "master_weight", | |
| "optimizer_state", | |
| "router_logits", | |
| "sinkhorn", | |
| "memory_attention_logits", | |
| "loss_accumulation", | |
| "mamba_sensitive" | |
| ], | |
| "fp8_format": "hybrid_e4m3_e5m2", | |
| "fp8_roles": [ | |
| "mamba_in_projection", | |
| "mamba_out_projection", | |
| "attention_qkv_projection", | |
| "attention_out_projection", | |
| "expert_gate_up_projection", | |
| "expert_down_projection", | |
| "latent_down_projection", | |
| "latent_up_projection", | |
| "memory_state_write_projection", | |
| "memory_query_projection", | |
| "memory_key_projection", | |
| "memory_value_projection", | |
| "memory_output_projection", | |
| "memory_route_projection", | |
| "mhc_controller", | |
| "ngram_projection", | |
| "lm_head" | |
| ], | |
| "fp8_scaling": "delayed", | |
| "master_weight_dtype": "float32", | |
| "optimizer_state_dtype": "float32", | |
| "parameter_dtype": "bfloat16", | |
| "reduction_dtype": "float32", | |
| "require_fp8_validation": true, | |
| "router_dtype": "float32" | |
| }, | |
| "reference_operating_point_flops_guard": false, | |
| "route_feature_dim": 256, | |
| "router_z_loss_coefficient": 0.001, | |
| "schema": "metis.model-family/v1", | |
| "sequence_length": 4096, | |
| "static_document_capacity": 0, | |
| "target_mean_passes": 2.0, | |
| "target_mean_routed_k": 4.0, | |
| "tie_embeddings": true, | |
| "vocab_size": 65536, | |
| "world_size": 1 | |
| } | |