| {"train/decoder_grad_norm": 0.21465465426445007, "train/decoder_param_norm": 10.98730182647705, "train/encoder_grad_norm": 5.377240449888632e-05, "train/encoder_param_norm": 21.97292137145996, "train/grad_norm": 0.21465465426445007, "layer_grad_norm/": {"decoder": {"model": {"decoder": {"embed_positions": {"embedding": 0.10678375512361526}, "embed_tokens": {"embedding": 0.18451564013957977}, "layernorm_embedding": {"bias": 0.006156071554869413, "scale": 0.0023730674292892218}, "layers": {"FlaxBartDecoderLayers": {"encoder_attn": {"k_proj": {"bias": 2.205506963040907e-13, "kernel": 6.429779668762348e-13}, "out_proj": {"bias": 0.00966663658618927, "kernel": 4.139078737352975e-05}, "q_proj": {"bias": 6.708300906953235e-13, "kernel": 6.82777673375462e-13}, "v_proj": {"bias": 0.0010709341149777174, "kernel": 3.5599548482423415e-06}}, "encoder_attn_layer_norm": {"bias": 0.009493962861597538, "scale": 0.003251356305554509}, "fc1": {"bias": 0.00018507803906686604, "kernel": 0.0003181225620210171}, "fc2": {"bias": 0.00927951280027628, "kernel": 0.0003024788456968963}, "final_layer_norm": {"bias": 0.010117009282112122, "scale": 0.0037409064825624228}, "self_attn": {"k_proj": {"bias": 8.830468023468541e-10, "kernel": 3.030844709428493e-06}, "out_proj": {"bias": 0.009151671081781387, "kernel": 0.0007765020127408206}, "q_proj": {"bias": 1.3603431625597295e-06, "kernel": 3.303162429801887e-06}, "v_proj": {"bias": 0.0007875036681070924, "kernel": 0.0007826802902854979}}, "self_attn_layer_norm": {"bias": 0.009495372883975506, "scale": 0.0032503989059478045}}}}}}, "encoder": {"adapter": {"layers": {"0": {"conv": {"bias": 2.578609326064907e-07, "kernel": 1.2006182714685565e-06}}, "1": {"conv": {"bias": 2.988664618897019e-06, "kernel": 1.2403276059558266e-06}}, "2": {"conv": {"bias": 5.35920298716519e-05, "kernel": 2.3276438696484547e-06}}}}, "encoder": {"layer_norm": {"bias": 3.93576407020646e-08, "scale": 3.677495286069643e-08}, "layers": {"FlaxWav2Vec2EncoderLayers": {"attention": {"k_proj": {"bias": 1.2308367046659063e-15, "kernel": 1.6772054578506612e-10}, "out_proj": {"bias": 8.37909226447664e-07, "kernel": 1.422491067160081e-07}, "q_proj": {"bias": 5.139029826684016e-11, "kernel": 1.5804688113796317e-10}, "v_proj": {"bias": 5.5300837686900195e-08, "kernel": 1.2766930979069002e-07}}, "feed_forward": {"intermediate_dense": {"bias": 3.932927228333938e-08, "kernel": 1.198425820803095e-07}, "output_dense": {"bias": 8.135644975482137e-07, "kernel": 1.2023470219446608e-07}}, "final_layer_norm": {"bias": 2.726392667184996e-09, "scale": 1.922166203982556e-09}, "layer_norm": {"bias": 4.49537962410318e-09, "scale": 2.7003910219036698e-09}}}, "pos_conv_embed": {"conv": {"bias": 2.595295711671497e-07, "weight_g": 5.773488354066103e-09, "weight_v": 6.550413189643223e-08}}}, "feature_extractor": {"conv_layers": {"0": {"conv": {"kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "1": {"conv": {"kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}, "2": {"conv": {"kernel": 0.0}, "layer_norm": {"bias": 0.0, "scale": 0.0}}}}, "feature_projection": {"layer_norm": {"bias": 4.7881744791311576e-08, "scale": 7.146309766170589e-09}, "projection": {"bias": 4.984443080502388e-07, "kernel": 4.582484223192296e-07}}, "masked_spec_embed": 0.0}}, "layer_param_norm/": {"decoder": {"model": {"decoder": {"embed_positions": {"embedding": 0.7995269894599915}, "embed_tokens": {"embedding": 2.509093999862671}, "layernorm_embedding": {"bias": 0.003853189293295145, "scale": 4.000099182128906}, "layers": {"FlaxBartDecoderLayers": {"encoder_attn": {"k_proj": {"bias": 2.311868696480701e-09, "kernel": 0.43237069249153137}, "out_proj": {"bias": 0.005428471136838198, "kernel": 0.4662071168422699}, "q_proj": {"bias": 2.3312416885801213e-08, "kernel": 0.4763868451118469}, "v_proj": {"bias": 0.005809006281197071, "kernel": 0.46249714493751526}}, "encoder_attn_layer_norm": {"bias": 0.005538478493690491, "scale": 5.656983852386475}, "fc1": {"bias": 0.0027363670524209738, "kernel": 0.23037242889404297}, "fc2": {"bias": 0.0054403552785515785, "kernel": 0.2286112755537033}, "final_layer_norm": {"bias": 0.005335222464054823, "scale": 5.657033920288086}, "self_attn": {"k_proj": {"bias": 1.2574997526826337e-05, "kernel": 0.44869107007980347}, "out_proj": {"bias": 0.005434151738882065, "kernel": 0.4594263732433319}, "q_proj": {"bias": 0.004717462230473757, "kernel": 0.43884801864624023}, "v_proj": {"bias": 0.005249769426882267, "kernel": 0.47378775477409363}}, "self_attn_layer_norm": {"bias": 0.005538613069802523, "scale": 5.656983852386475}}}}}}, "encoder": {"adapter": {"layers": {"0": {"conv": {"bias": 0.003248997498303652, "kernel": 0.7977768182754517}}, "1": {"conv": {"bias": 0.004121119622141123, "kernel": 0.7821999788284302}}, "2": {"conv": {"bias": 0.0039659361355006695, "kernel": 0.793234646320343}}}}, "encoder": {"layer_norm": {"bias": 0.0011272807605564594, "scale": 4.000342845916748}, "layers": {"FlaxWav2Vec2EncoderLayers": {"attention": {"k_proj": {"bias": 1.1786402409619257e-10, "kernel": 0.6511668562889099}, "out_proj": {"bias": 0.003823851700872183, "kernel": 0.6514734625816345}, "q_proj": {"bias": 8.803545824775938e-06, "kernel": 0.6364570260047913}, "v_proj": {"bias": 0.002547932555899024, "kernel": 0.6420494318008423}}, "feed_forward": {"intermediate_dense": {"bias": 0.002046645386144519, "kernel": 0.7063329219818115}, "output_dense": {"bias": 0.004090356640517712, "kernel": 0.7153141498565674}}, "final_layer_norm": {"bias": 0.0002977659460157156, "scale": 7.99997615814209}, "layer_norm": {"bias": 0.0005923936259932816, "scale": 8.000036239624023}}}, "pos_conv_embed": {"conv": {"bias": 0.001659030676819384, "weight_g": 2.2776856422424316, "weight_v": 2.277585744857788}}}, "feature_extractor": {"conv_layers": {"0": {"conv": {"kernel": 7.867799758911133}, "layer_norm": {"bias": 0.0, "scale": 5.656854152679443}}, "1": {"conv": {"kernel": 8.025212287902832}, "layer_norm": {"bias": 0.0, "scale": 5.656854152679443}}, "2": {"conv": {"kernel": 7.975250720977783}, "layer_norm": {"bias": 0.0, "scale": 5.656854152679443}}}}, "feature_projection": {"layer_norm": {"bias": 0.0014101800043135881, "scale": 5.656826972961426}, "projection": {"bias": 0.0016126643167808652, "kernel": 0.4301906228065491}}, "masked_spec_embed": 2.404470205307007}}, "train/learning_rate": 2.400018274784088e-06, "train/loss": 6.901019096374512, "train/param_norm": 24.566848754882812, "_timestamp": 1651945950, "_runtime": 69, "_step": 4} |