Update README.md
Browse files
README.md
CHANGED
|
@@ -29,7 +29,7 @@ DynaMix is based on a sparse mixture of experts (MoE) architecture operating in
|
|
| 29 |
|
| 30 |
By aggregating the expert weighting with the expert prediction the next state is predicted. The current model has the following specifics:
|
| 31 |
|
| 32 |
-
- **M (Latent state dimension):**
|
| 33 |
- **N (Observation space dimension):** 3
|
| 34 |
- **Experts:** 10 expert networks in the mixture
|
| 35 |
- **Expert type:** `"almost_linear_rnn"` — a compact recurrent model combining linear and nonlinear components (`P=2` ReLU units)
|
|
@@ -50,7 +50,7 @@ model = DynaMix(M=M, N=N, Experts=EXPERTS, expert_type=EXPERT_TYPE, P=P)
|
|
| 50 |
# Load model weights
|
| 51 |
model_path = hf_hub_download(
|
| 52 |
repo_id="DurstewitzLab/dynamix-3d",
|
| 53 |
-
filename="dynamix-3d-
|
| 54 |
)
|
| 55 |
model_state_dict = load_file(model_path)
|
| 56 |
model.load_state_dict(model_state_dict)
|
|
|
|
| 29 |
|
| 30 |
By aggregating the expert weighting with the expert prediction the next state is predicted. The current model has the following specifics:
|
| 31 |
|
| 32 |
+
- **M (Latent state dimension):** 30
|
| 33 |
- **N (Observation space dimension):** 3
|
| 34 |
- **Experts:** 10 expert networks in the mixture
|
| 35 |
- **Expert type:** `"almost_linear_rnn"` — a compact recurrent model combining linear and nonlinear components (`P=2` ReLU units)
|
|
|
|
| 50 |
# Load model weights
|
| 51 |
model_path = hf_hub_download(
|
| 52 |
repo_id="DurstewitzLab/dynamix-3d",
|
| 53 |
+
filename="dynamix-3d-base-v1.0.safetensors"
|
| 54 |
)
|
| 55 |
model_state_dict = load_file(model_path)
|
| 56 |
model.load_state_dict(model_state_dict)
|