Skip to content

Cleanup modeling file for Apriel-H #379

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 5 commits into
base: hybrid_dev
Choose a base branch
from
Open

Cleanup modeling file for Apriel-H #379

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
f8de21f
change m2_indexes to multiple
nitsanluke Sep 29, 2025
1a1e4b9
Merge branch 'hybrid_dev' of github.com:ServiceNow/Fast-LLM into hybr…
nitsanluke Oct 16, 2025
ed800ae
formatting
nitsanluke Oct 16, 2025
10fdd92
drop wrong file
nitsanluke Oct 16, 2025
bbec888
reset file
nitsanluke Oct 16, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion fast_llm/models/ssm/external/apriel_15b_hybrid/configuration_ssm_hybrid_apriel15b.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,7 @@
class AprielSSMHybridConfig(MistralConfig):
model_type = "apriel_ssm_thinker_hybrid"

def __init__(self, hybrid_block_layout=["m2d"], ssm_cfg=None, **kwargs):
Copy link
Contributor Author

@nitsanluke nitsanluke Oct 16, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should we clean up this configs? @oleksost

ssm_config_default = {	ssm_config_default = {
    # discrete mamba2	    # discrete mamba2
    "d_state": 64,	    "d_state": 64,
    "n_v_heads": 32,	    "n_v_heads": 32,
    "n_qk_heads": 32,	    "n_qk_heads": 32,
    "expand": 1,	    "expand": 1,
    "chunk_size": 128,	    "chunk_size": 128,
    "activation": "identity",	    "activation": "identity",
    "bias": False,	    "bias": False,
    "d_conv": 4,	    "d_conv": 4,
    "d_inner": 32 * 128,	    "d_inner": 32 * 128,
    # mamba2	    # mamba2
    "d_xb": None,  # will be set to model dim	    "d_xb": None,  # will be set to model dim
    "dt_rank": "auto",	    "dt_rank": "auto",
    "dt_min": 0.001,	    "dt_min": 0.001,
    "dt_max": 0.1,	    "dt_max": 0.1,
    "dt_init": "random",	    "dt_init": "random",
    "dt_scale": 1.0,	    "dt_scale": 1.0,
    "dt_init_floor": 1e-4,	    "dt_init_floor": 1e-4,
    "conv_bias": True,
    ```

def __init__(self, hybrid_block_layout=["m2"], ssm_cfg=None, **kwargs):
super().__init__(**kwargs)
self.hybrid_block_layout = hybrid_block_layout
self.head_dim = self.head_dim or self.hidden_size // self.num_attention_heads # as in transformers 4.51.3
Expand Down
Loading