Here's something from two references I found.
The reduction is carried out in closed porous cells, with lead electrodes, and with 20% sulfuric acid as the anode liquid. The cathode solution is prepared by dissolving the amide in a mixture of alcohol, water and sulfuric acid. Formamide, acetamide and N,N-dimethylpropionamide are very resistant to reduction, but the N-substituted amides reduce more readily than do the parent compounds.88
88 J. Gen. Chem. 11, 51, (1941)
from An Outline of Organic Nitrogen Compounds by E. F. Degering 1950
Reduction Reduction of amides has been mentioned in Chapter 2 as a useful preparative method for amines. Mild reducing agents do not affect amides. Electrolytic reduction,24 usually with lead or zinc amalgam electrodes in sulfuric acid solution, used to be the best method, although it is generally successful only with amides of aromatic acids and with dialkylamides of aliphatic acids. Hydrogenation can also be useful, but the general inertness of amides makes high pressures and temperatures necessary. Recent studies 25 have shown rhenium to be an unusually effective catalyst, however. Sodium and hot alcohol, a reducing system of wide application, is of very limited value with amides. Many amides are little affected, and those that do react undergo C-N cleavage to a major extent, producing the alcohol derived from the acyl group. So-called Birch reduction-sodium, calcium, etc., in liquid ammonia plus a proton source, such as alcohols-produces aldehydes, 26 and serves as a useful deacylating method.
The metal hydrides, essentially lithium aluminum hydride and its derivatives, are the foremost reducing agents for amides. With proper choice of reagent and proportions, either amines or aldehydes can be obtained in preparatively useful yields (Eq. 40). Reduction apparently
goes in two stages, first by addition of hydride to the carbonyl group, forming a carbinolamine (as its aluminate complex), which can be hydrolyzed to an aldehyde. In the next stage, either a C-N or a C-O bond must be broken, resulting respectively in an alcohol, or an amine. Of course, it is only when the first stage is faster than the second that it is possible to stop the reduction at the carbinolamine stage for the purpose of aldehyde synthesis. Such conditions prevail with N-methylanilides in general when lithium aluminum hydride is used,27 and with dialkylamides when lithium diethoxyaluminum hydride, LiAlH2(OC2H5)2, is the reducing agent 28 (Eqs. 41, 42). Anhydrous aluminum chloride
used with lithium aluminum hydride greatly favors C-O cleavage in the second reduction stage, and thus improves the conversion of amides to amines29 (Eq. 42). Unsubstituted amides appear to be dehydrated in the first stage of reaction with lithium aluminum hydride, giving nitriles, which are, of course, reduced rapidly to amines.30 With such a reaction path, aldehyde formation of the type described above cannot occur, and yields of amines are accordingly high.
from The Chemistry of Open Chain Organic Nitrogen Compounds, Vol. 1, P. A. S. Smith, 1965
http://www.geocities.com/dritte123/PSPF.htmlThe hardest thing to explain is the obvious
