Skip to content
GitHub
View on GitHub

Qwen3-14B

Qwen3-14B (14 billion parameters) from Alibaba.

from modal_training_gym.common.models.qwen3_14b import Qwen3_14B

Qwen3-14B (14 billion parameters) from Alibaba.

Inherits from: HFModelConfiguration, ModelConfig

FieldTypeDefaultDescription
model_namestr"Qwen/Qwen3-14B"
model_pathstr | NoneNone
architectureModelArchitecture | NoneModelArchitecture(num_layers=40, hidden_size=5120, ffn_hidden_size=17408, num_attention_heads=40, group_query_attention=True, num_query_groups=8, kv_channels=128, vocab_size=151936, normalization='RMSNorm', norm_epsilon=1e-06, swiglu=True, disable_bias_linear=True, qk_layernorm=True, untie_embeddings_and_output_weights=True, num_experts=0, moe_ffn_hidden_size=0, moe_shared_expert_intermediate_size=0, moe_grouped_gemm=False, moe_shared_expert_gate=False, moe_router_topk=0, moe_router_score_function='', moe_token_drop_policy='', moe_router_dtype='', moe_permute_fusion=False, moe_aux_loss_coeff=None, megatron_spec=None, megatron_model_type='', apply_layernorm_1p=False, use_gated_attention=False, attention_output_gate=False, use_rotary_position_embeddings=True, rotary_base=1000000, rotary_percent=1.0)
response_parserOptional[Callable[[str], ParsedResponse]]<function parse_qwen3_response at 0x103523240>

Download or materialize weights into the model volume.

parse_response(self, text: 'str') -> 'ParsedResponse'

Section titled “parse_response(self, text: 'str') -> 'ParsedResponse'”

Parse raw model output into structured content.

response_parser(text: 'str') -> 'ParsedResponse'

Section titled “response_parser(text: 'str') -> 'ParsedResponse'”

Parse Qwen3-family model output into structured content.

Source: modal_training_gym/common/models/qwen3_14b.py