Instructions to use mkurman/convgpt-v2-b200-full-synth-20h with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use mkurman/convgpt-v2-b200-full-synth-20h with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="mkurman/convgpt-v2-b200-full-synth-20h")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("mkurman/convgpt-v2-b200-full-synth-20h", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use mkurman/convgpt-v2-b200-full-synth-20h with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "mkurman/convgpt-v2-b200-full-synth-20h" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "mkurman/convgpt-v2-b200-full-synth-20h", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/mkurman/convgpt-v2-b200-full-synth-20h
- SGLang
How to use mkurman/convgpt-v2-b200-full-synth-20h 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 "mkurman/convgpt-v2-b200-full-synth-20h" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "mkurman/convgpt-v2-b200-full-synth-20h", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "mkurman/convgpt-v2-b200-full-synth-20h" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "mkurman/convgpt-v2-b200-full-synth-20h", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use mkurman/convgpt-v2-b200-full-synth-20h with Docker Model Runner:
docker model run hf.co/mkurman/convgpt-v2-b200-full-synth-20h
| { | |
| "architectures": [ | |
| "ConvGPTV2ForCausalLM" | |
| ], | |
| "attention_every": 0, | |
| "attention_gate_init": -2.0, | |
| "attention_num_heads": 4, | |
| "attention_start_layer": 0, | |
| "bos_token_id": 1, | |
| "branch_dropout": 0.0, | |
| "chunk_memory_gate_init": -4.0, | |
| "chunk_memory_include_current_chunk": false, | |
| "chunk_memory_size": 64, | |
| "chunk_memory_token_top_k": 64, | |
| "chunk_memory_top_k": 2, | |
| "conv1d_dilations": [ | |
| 1, | |
| 2, | |
| 4, | |
| 8 | |
| ], | |
| "conv1d_kernel_size": 3, | |
| "conv1d_residual_gate_init": -6.0, | |
| "conv2d_backend": "triton_gather", | |
| "conv2d_chunk_size": 1024, | |
| "conv2d_dilations": [ | |
| 1, | |
| 2, | |
| 4, | |
| 8 | |
| ], | |
| "conv2d_kernel_size": 3, | |
| "conv2d_residual_gate_init": 2.0, | |
| "conv_expand": 2, | |
| "dropout": 0.0, | |
| "dtype": "bfloat16", | |
| "eos_token_id": 32009, | |
| "fusion": "gated", | |
| "gated_delta_conv_kernel_size": 4, | |
| "gated_delta_every": 0, | |
| "gated_delta_gate_init": 0.0, | |
| "gated_delta_key_dim": 64, | |
| "gated_delta_num_heads": 4, | |
| "gated_delta_start_layer": 0, | |
| "gated_delta_value_dim": 64, | |
| "grid_size": 256, | |
| "hidden_act": "silu", | |
| "hidden_size": 512, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 2048, | |
| "linear_attention_every": 0, | |
| "linear_attention_gate_init": 0.0, | |
| "linear_attention_num_heads": 4, | |
| "linear_attention_start_layer": 0, | |
| "max_grid_tokens": 65536, | |
| "max_position_embeddings": 65536, | |
| "model_type": "convgpt_v2", | |
| "num_hidden_layers": 32, | |
| "pack_order": "sequence_to_curve", | |
| "packing": "hilbert", | |
| "pad_token_id": 32000, | |
| "position_embedding_type": "rope_nope", | |
| "retrieval_dropout": 0.0, | |
| "retrieval_every": 2, | |
| "retrieval_num_heads": 4, | |
| "retrieval_num_slots": 64, | |
| "retrieval_top_k": 4, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "router_rope_fraction": 1.0, | |
| "router_type": "chunk_token_memory", | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.9.0", | |
| "two_d_every": 1, | |
| "two_d_start_layer": 0, | |
| "use_1d_branch": true, | |
| "use_2d_branch": true, | |
| "use_2d_depthwise": true, | |
| "use_cache": false, | |
| "use_row_col_embeddings": false, | |
| "vocab_size": 32024 | |
| } | |