Skip to content

Add support for Mistral-Nemo-Base-2407 model (#751) #1

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
frances720 wants to merge 1 commit into
base: frances/test_mistral
Choose a base branch
from
Open

Add support for Mistral-Nemo-Base-2407 model (#751) #1

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
3 changes: 2 additions & 1 deletion demos/Colab_Compatibility.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"TransformerLens currently supports 185 models out of the box.\n"
"TransformerLens currently supports 186 models out of the box.\n"
]
}
],
Expand Down Expand Up @@ -331,6 +331,7 @@
" \"microsoft/Phi-3-mini-4k-instruct\",\n",
" \"mistralai/Mistral-7B-Instruct-v0.1\",\n",
" \"mistralai/Mistral-7B-v0.1\",\n",
" \"mistralai/Mistral-Nemo-Base-2407\",\n",
" \"Qwen/Qwen-7B\",\n",
" \"Qwen/Qwen-7B-Chat\",\n",
" \"Qwen/Qwen1.5-4B\",\n",
Expand Down
29 changes: 16 additions & 13 deletions transformer_lens/loading_from_pretrained.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -177,6 +177,7 @@
"stabilityai/stablelm-tuned-alpha-7b",
"mistralai/Mistral-7B-v0.1",
"mistralai/Mistral-7B-Instruct-v0.1",
"mistralai/Mistral-Nemo-Base-2407",
"mistralai/Mixtral-8x7B-v0.1",
"mistralai/Mixtral-8x7B-Instruct-v0.1",
"bigscience/bloom-560m",
Expand Down Expand Up @@ -604,6 +605,7 @@
],
"mistralai/Mistral-7B-v0.1": ["mistral-7b"],
"mistralai/Mistral-7B-Instruct-v0.1": ["mistral-7b-instruct"],
"mistralai/Mistral-Nemo-Base-2407": ["mistral-nemo-base-2407"],
"mistralai/Mixtral-8x7B-v0.1": ["mixtral", "mixtral-8x7b"],
"mistralai/Mixtral-8x7B-Instruct-v0.1": [
"mixtral-instruct",
Expand Down Expand Up @@ -1070,24 +1072,25 @@ def convert_hf_model_config(model_name: str, **kwargs):
"attention_dir": "bidirectional",
}
elif architecture == "MistralForCausalLM":
use_local_attn = True if hf_config.sliding_window else False
cfg_dict = {
"d_model": 4096,
"d_head": 4096 // 32,
"n_heads": 32,
"d_mlp": 14336,
"n_layers": 32,
"d_model": hf_config.hidden_size,
"d_head": hf_config.head_dim or hf_config.hidden_size // hf_config.num_attention_heads,
"n_heads": hf_config.num_attention_heads,
"d_mlp": hf_config.intermediate_size,
"n_layers": hf_config.num_hidden_layers,
"n_ctx": 2048, # Capped due to memory issues
"d_vocab": 32000,
"act_fn": "silu",
"d_vocab": hf_config.vocab_size,
"act_fn": hf_config.hidden_act,
"window_size": hf_config.sliding_window, # None if no sliding window was used
"attn_types": ["local"] * hf_config.num_hidden_layers if use_local_attn else None,
"eps": hf_config.rms_norm_eps,
"rotary_base": hf_config.rope_theta,
"n_key_value_heads": hf_config.num_key_value_heads,
"use_local_attn": use_local_attn,
"normalization_type": "RMS",
"positional_embedding_type": "rotary",
"window_size": 4096,
"attn_types": ["local"] * 32,
"eps": 1e-05,
"n_key_value_heads": 8,
"gated_mlp": True,
"use_local_attn": True,
"rotary_dim": 4096 // 32,
}
elif architecture == "MixtralForCausalLM":
cfg_dict = {
Expand Down